Microneedle-Based Optical Biosensors for Point-of-Care Glucose Monitoring in Interstitial Fluid: Recent Advances, Current Challenges, and Future Perspectives

Abstract

Microneedle (MN)-based optical biosensors are emerging as minimally invasive platforms for point-of-care testing (POCT) and wearable glucose monitoring by enabling direct access to interstitial fluid (ISF), which closely reflects plasma glucose dynamics. This review summarizes recent advances in MN-based glucose sensing with particular emphasis on optical biosensors. Blood-to-ISF glucose transport, physiological time lag, MN architectures, fabrication strategies, and ISF sampling mechanisms are discussed. Glucose recognition mechanisms are outlined together with optical transduction approaches, including colorimetric, fluorescence, surface-enhanced Raman scattering (SERS), photonic, and plasmonic sensing. Many colorimetric MN platforms currently function as single-use or endpoint POCT systems, whereas reversible fluorescence- and photonic-based sensors provide greater potential for repeated measurements and continuous glucose monitoring. Advanced functional materials, smartphone-assisted readout, and emerging artificial intelligence/machine-learning approaches for signal processing, calibration, and glucose prediction are also examined. Key analytical, mechanical, biocompatibility, and ex vivo/in vivo performance parameters are critically evaluated, together with patents, clinical investigations, and commercialization status. Compared with previous reviews that mainly address ISF sampling, general MN biosensing, glucose monitoring across multiple sensing modalities, or wearable optical sensors separately. This review provides an integrated glucose-specific framework linking ISF physiology, MN-mediated sampling, point-of-care and emerging continuous optical sensing, advanced materials, digital/AI analysis, performance validation, and clinical translation. Finally, current limitations involving biofouling, signal drift, long-term stability, manufacturing scalability, and clinical validation are discussed to highlight future directions toward advanced point-of-care platforms and emerging intelligent, wearable, and continuous glucose-monitoring systems.

1. Introduction

Diabetes mellitus (DM) describes a group of chronic and metabolic diseases characterized by the dysregulation of glucose homeostasis, resulting in high levels of blood glucose (hyperglycemia). It is also a remarkably prevalent and challenging health problem worldwide. As reported by the International Diabetes Federation (IDF), the global population of adults aged 20–60 with diabetes was approximately 537 million [1], accounting for 10.5% of the global adult population. This condition causes approximately 1.5 million people to die yearly [2]. DM has become a main cause of several health issues such as cardiovascular disease, kidney failure, blindness, stroke, and limb amputation in many countries. In recent years, point-of-care testing (POCT) has become a transformative tool and has gained significant attention in advancing proactive healthcare by offering simplicity, affordability, and accessibility for clinical applications [3]. These devices enable rapid disease monitoring and screening. Monitoring blood-glucose levels is a core component of managing diabetes and diagnosing disorders associated with glucose metabolism [4]. Glucose monitoring techniques include invasive finger-prick testing, minimally invasive continuous glucose monitoring (CGM) systems, and emerging non-invasive optical or biochemical approaches [5]. These methods aim to improve diabetes management by enhancing patient comfort, reducing the burden of frequent blood sampling, and providing accurate, real-time glucose data for better glycemic control.

Blood tests have long been considered the common standard for clinical diagnosis. Traditional blood-glucose level testing typically requires invasive blood collection through a fingerstick, which frequently results in pain, potential a risk of cross-contamination, low patient adherence, and inconvenience [6,7]. These have become issues that often contribute to trypanophobia among diabetic patients. Furthermore, it fails to satisfy the need for real-time blood-glucose monitoring due to the requirement of multiple daily punctures, and the absence of critical physiological signals can lead to delays in medical action. Currently, novel or alternative approaches for glucose monitoring have been developed to overcome the limitations of traditional blood-glucose monitoring. These approaches shift from invasive methods to minimally invasive and non-invasive techniques, including non-invasive optical methods [8,9] and sampling from sites other than the fingertip [10,11,12]. For minimally invasive applications, microneedles (MNs) biosensors offer a transformative and patient-friendly diagnostic platform. They integrate the minimally painful penetration of MNs with the immediate feedback, enabling simple and rapid detection. By accessing interstitial fluid (ISF) beneath the skin surface, these devices enable real-time, in situ analysis and provide a practical alternative to conventional blood-based testing methods [13].

Glucose biosensors generally consist of a biorecognition element coupled with a transducer that converts glucose-dependent biochemical interactions into measurable signals. Glucose recognition is commonly achieved through enzymatic systems, particularly glucose oxidase (GOx) and glucose dehydrogenase (GDH), or through non-enzymatic strategies using catalytic materials and synthetic receptors. The resulting biochemical response can be converted through electrochemical or optical transduction mechanisms. Electrochemical glucose biosensors remain the most established approach because of their high sensitivity, rapid response, and compatibility with miniaturized electronics, whereas optical biosensors provide complementary advantages, including resistance to electromagnetic interference, remote signal acquisition, multiplexing capability, and compatibility with fluorescence, colorimetric, Raman, photonic, and plasmonic readouts. These sensing principles provide the fundamental basis for the development of minimally invasive glucose-monitoring platforms [12].

Wearable biosensors extend these principles by integrating sensing components with flexible substrates, miniaturized electronics, wireless communication, and digital interfaces to enable continuous or near-continuous monitoring under real-life conditions. For glucose monitoring, successful wearable systems require not only reliable molecular recognition but also efficient access to biofluids, stable signal acquisition, low-power operation, and robust calibration during prolonged use [14]. Microfluidic technologies are particularly important in this context because they can regulate the collection, transport, metering, and delivery of small-volume biofluids to sensing regions while minimizing sample loss and contamination [15]. When integrated with MNs, microfluidic channels and capillary-driven structures can facilitate ISF transport from the skin toward optical or electrochemical sensing elements. While hollow, porous, and hydrogel-forming MNs can further enhance fluid access through pressure-driven flow, capillary action, or swelling-assisted extraction [16,17]. Therefore, the convergence of glucose biosensing, wearable electronics, microfluidics, and MNs-mediated ISF access provides a technological foundation for next-generation minimally invasive glucose-monitoring systems.

An important distinction should be made between continuous glucose monitoring and discrete MN-based glucose measurements. Continuous glucose monitoring refers to repeated or automated glucose measurements acquired over an extended period while the sensing device remains in place, thereby providing temporal glucose profiles, trends, and rates of change. In contrast, intermittent monitoring involves repeated but discrete sampling or measurements at selected time intervals, often requiring separate MNs application or ISF collection for each analysis. Single-point measurement represents a single glucose determination obtained from one sampling event and does not provide information on temporal glucose dynamics. Accordingly, MNs platforms that only extract ISF followed by off-device analysis or provide a single colorimetric, fluorescence, or SERS measurement should not be classified as continuous glucose-monitoring systems unless repeated longitudinal measurements have been experimentally demonstrated.

This narrative review provides an integrated overview of recent advances in MNs-based biosensors for glucose monitoring in ISF. It focuses on the fundamental relationship between blood and ISF glucose, MNs design and fabrication strategies, ISF sampling mechanisms, and the integration of MNs with electrochemical and optical sensing platforms. Particular emphasis is placed on emerging sensing approaches, including fluorescence, colorimetric, surface-enhanced Raman scattering (SERS), nanomaterial-assisted detection, and wearable biosensor systems. The review also evaluates key analytical performance parameters, such as sensitivity, selectivity, response time, stability, reproducibility, and resistance to biological interference. In addition, recent progress in artificial intelligence-assisted signal processing, calibration, glucose trend prediction, and personalized monitoring is discussed. Clinical translation, patents, commercialization, and regulatory considerations are further summarized to highlight the gap between laboratory-scale prototypes and practical glucose-monitoring devices. Finally, current challenges related to ISF extraction efficiency, signal drift, biofouling, long-term stability, manufacturing scalability, and clinical validation are critically examined, together with future perspectives toward intelligent, minimally invasive, and continuous glucose-monitoring systems.

Although several reviews have addressed MNs-mediated ISF sampling, MNs-based glucose monitoring, general MN biosensors, and wearable optical MN sensors, the available literature remains fragmented across these individual domains. The present review provides a distinct synthesis by specifically linking blood-to-ISF glucose physiology and physiological time lag with MNs architecture, ISF sampling mechanisms, glucose-recognition chemistry, and optical signal transduction. Particular emphasis is placed on colorimetric, fluorescence, SERS, photonic, and plasmonic MNs platforms and the advanced functional materials that enable these sensing mechanisms. In addition, this review integrates smartphone-assisted readout and emerging AI/ML approaches for optical image processing, spectral analysis, multivariable calibration, interference correction, and glucose prediction. Importantly, analytical and mechanical performance, ex vivo/in vivo validation, biocompatibility, patents, clinical investigations, and commercialization are evaluated within the same glucose-specific framework. Thus, rather than providing another general overview of MNs biosensors, this review establishes a translational pathway from ISF glucose physiology and MN-mediated sensing to intelligent optical glucose-monitoring systems. Comparison of the present review with previous reviews on MNs-based biosensing and glucose monitoring was represented in Table 1.

Table 1. Comparison of the present review with previous reviews on MNs-based biosensing and glucose monitoring.

Review	Main Scope	Glucose and ISF Focus	Optical MN Coverage	AI/Digital Analysis	Translation/Clinical Perspective	Distinction from the Present Review
Saifullah and Faraji Rad, 2023 [18]	MN-mediated dermal ISF sampling, extraction mechanisms, sampling volume/duration, and biosensor integration	ISF is the primary biofluid; glucose is one of several biomarkers	Optical sensing is included within broader biosensing applications	Not a central focus	Future opportunities discussed	Primarily organized around ISF extraction and sampling, rather than glucose-specific optical transduction and data interpretation.
Wang et al., 2023 [14]	MN-based glucose sampling and wearable glucose biosensors	Glucose-specific; detailed MN sampling mechanisms	Raman, colorimetric, fluorescence, and electrochemical sensing are included	Not a major organizing theme	Future wearable development discussed	Closest glucose-focused predecessor, but treats electrochemical and optical approaches together; it does not organize the field around optical glucose MNs, advanced optical materials, AI-assisted optical analysis, and an updated translational landscape.
Kim et al., 2024 [19]	MN sensors for dermal ISF analysis, including design, fabrication, sensing mechanisms, monitoring electronics, and applications	Strong ISF focus; glucose is one of multiple biomarkers	Covers fluorescence, colorimetric, and plasmonic/SERS sensing alongside electrochemical approaches	Wearable monitoring electronics are discussed; AI is mentioned mainly as a future direction	Discusses wearable integration, clinical translation, scalable fabrication, and future personalized healthcare	Broader focus on multi-biomarker ISF sensing and multiple transduction modes; the present review specifically emphasizes optical glucose sensing, blood–ISF physiology, POCT/CGM, advanced materials, AI-assisted analysis, and clinical translation.
Hu et al., 2024 [20]	MN sensors for point-of-care diagnostics and personalized medicine	Multiple analytes in ISF	Electrochemical and optical sensing discussed	Intelligent/POC concepts discussed broadly	POC and personalized-medicine perspectives	Focuses on “lab-on-a-MNs” diagnostics across biomarkers rather than detailed comparison of colorimetric, fluorescence, SERS, photonic, and plasmonic glucose sensing.
Yang et al., 2025 [21]	MN-assisted SERS sampling and detection	Broad molecular detection; glucose is an application rather than the exclusive focus	Strong SERS focus	Not a central focus	Challenges and future development discussed	Provides detailed SERS coverage but does not integrate SERS with other optical glucose modalities or blood–ISF physiology, AI, and clinical translation.
Xiao et al., 2025 [22]	Wearable optical MN sensors for healthcare monitoring	Multiple biomarkers rather than glucose alone	Strong focus on colorimetry, fluorescence, and SERS; includes fabrication and materials	Digital readout is relevant, but AI-based glucose analysis is not the principal framework	Challenges and future directions discussed	Strong optical overlap, but broader in analyte scope. The present review narrows the analysis to glucose in ISF and links sensing mechanisms to glucose physiology, quantitative performance, AI/ML, patents, and commercialization.
Zhao et al., 2025 [15]	Engineering MNs for biosensing and drug delivery, including materials, fabrication, and clinical translation	Broad biofluid sensing; not glucose-specific	Fluorescence, SERS, colorimetric, and electrical readouts are covered	Not a central focus	Clinical translation and commercialization are important components	Broader engineering and drug-delivery review; does not provide a dedicated optical glucose-monitoring framework.
Sun et al., 2026 [23]	Optical MNs patches for point-of-care analysis and theranostics	Glucose included among several biomarkers	Strong focus on photonic crystal, fluorescent, and plasmonic MNs	Not a principal focus	Clinical translation/commercialization challenges discussed generally	Focuses on optical MN design and POC applications broadly; the present review specifically synthesizes ISF glucose monitoring, including SERS/colorimetry, glucose recognition chemistry, AI-assisted quantification, and glucose-specific validation.
Cho et al., 2026 [24]	Wearable MN sensors for continuous ISF monitoring	ISF and continuous monitoring are central; multiple clinical applications	Primarily electrochemical	Digital/closed-loop integration considered	Strong regulatory, workflow, and translational focus	Complements rather than overlaps with the present review because it emphasizes long-term electrochemical ISF monitoring rather than optical glucose sensing.
Present review	MN-based optical biosensors specifically for glucose monitoring in ISF	Dedicated focus on blood–ISF glucose transport, physiological lag, ISF variability, sampling, and glucose detection	Colorimetry, fluorescence, SERS, photonic, and plasmonic sensing, with electrochemical systems retained mainly as a comparator	Smartphone readout, optical image/spectral processing, ML calibration, SERS analysis, signal correction, and glucose prediction	Performance validation, biocompatibility, patents, clinical investigations, commercialization, and benchmarking against commercial CGMs	Integrates the complete pathway from blood-to-ISF physiology → MN access/sampling → glucose recognition → optical transduction/materials → digital/AI analysis → analytical/in vivo validation → clinical translation within one glucose-specific framework.

This article was conducted as a narrative review to provide an integrated overview of microneedle-based optical biosensors for glucose monitoring in ISF. Relevant literature was identified through searches of PubMed, Scopus, Web of Science, and Google Scholar, focusing primarily on publications from 2015 to 2026, while earlier seminal studies were included when necessary to describe fundamental concepts and technological development. Search terms included combinations of “microneedle”, “glucose biosensor”, “interstitial fluid”, “optical biosensor”, “colorimetric”, “fluorescence”, “SERS”, “photonic”, “plasmonic”, “continuous glucose monitoring”, “wearable biosensor”, and “artificial intelligence”. Priority was given to original research articles describing MNs-mediated ISF sampling, glucose-sensing mechanisms, optical transduction, analytical performance, in vitro/ex vivo/in vivo validation, and translational development. Relevant review articles, patents, clinical-trial records, and commercial-device information were additionally considered to provide technological and clinical context.

2. Interstitial Fluid (ISF) for Glucose Monitoring

Alternative biological samples for blood-glucose (BG) monitoring include sweat [25], saliva [26], tears [26], urine [26], and ISF [27]. Among these biofluids, ISF has emerged as a particularly suitable source for biomarker analysis because of its close compositional similarity to plasma, the presence of distinct biomarkers that may not be readily detectable in plasma, and its abundance within the extracellular compartment. ISF accounts for approximately 75% of extracellular fluid [28] and 15–25% of body weight [29]. It occupies the interstitial spaces between cells and is primarily derived from plasma that extravasates through capillary walls [30]. ISF contains water, electrolytes, nutrients, dissolved gases, metabolites, and other essential molecules required for cellular function. Accordingly, it serves as an immediate medium for nutrient delivery, waste removal, maintenance of cellular hydration, and intercellular communication [31,32]. Importantly, glucose concentrations in ISF are strongly correlated with those in plasma, supporting its use as an alternative fluid for glucose monitoring. For non-diabetic individuals, ISF glucose concentrations have been proposed to fall within approximately 70–140 mg/dL (3.9–7.8 mmol/L). In patients with type 1 diabetes (T1D) or type 2 diabetes (T2D), clinical recommendations for ISF-based continuous glucose monitoring (CGM) generally target a glucose range of 70–180 mg/dL for at least 70% of the monitored period [33].

Glucose is transported from the vascular compartment into the interstitial space across the capillary endothelium, after which it diffuses through the extracellular matrix before being taken up by surrounding cells. Therefore, ISF glucose concentration is determined by a dynamic balance between capillary glucose delivery, endothelial transport, interstitial diffusion, and local cellular-glucose consumption. Because equilibration between blood and ISF is not instantaneous, changes in ISF glucose generally follow changes in blood glucose with a physiological time lag of approximately 5–10 min, although the magnitude of this delay is variable rather than fixed. The lag may become more pronounced during rapid glucose excursions, such as after meals, insulin administration, exercise, or hypoglycemia, because the rate of glucose appearance or disappearance in blood may temporarily exceed the rate of equilibration with the interstitial compartment [34]. Local tissue perfusion and capillary permeability also influence glucose delivery to ISF, whereas cellular uptake and metabolic activity affect glucose removal from the interstitial space. Consequently, factors such as blood flow, exercise, temperature, hydration status, anatomical site, tissue composition, and the direction and rate of glucose change can alter the relationship between blood and ISF glucose concentrations [35]. For MN-based biosensors, additional local factors, including insertion depth, tissue compression, inflammation, ISF availability, and biofouling, may further influence glucose transport and sensing performance. Importantly, this physiological blood-to-ISF lag should be distinguished from sensor-related delay, which may arise from glucose diffusion through membranes or hydrogels, enzymatic reaction kinetics, optical or electrochemical signal generation, data acquisition, calibration, and algorithmic filtering. Thus, the total delay observed by an MN glucose sensor may represent the combined effects of physiological transport and device-specific response kinetics [36]. Glucose transport from the blood circulation into ISF, the physiological blood–ISF glucose time lag, and MNs-mediated access to dermal ISF for glucose sensing are illustrated in Figure 1.

Figure 1. Schematic illustration of glucose transport from the blood circulation into ISF, the physiological blood–ISF glucose time lag, and MNs-mediated access to dermal ISF for glucose sensing.

Despite these advantages, the broader utilization of ISF as a diagnostic fluid remains limited by the lack of rapid, minimally invasive techniques capable of obtaining sufficient sample volumes, typically approximately 1–10 µL [30,37]. Conventional ISF sampling approaches include suction blistering, sonophoresis, reverse iontophoresis, microdialysis, and thermal ablation [38]. However, some of these methods can disrupt the stratum corneum, induce local tissue damage, require prolonged sampling periods, or depend on specialized equipment, calibration, and trained personnel. These limitations restrict their practicality for routine point-of-care and self-monitoring applications. Therefore, the development of rapid, reliable, and minimally invasive ISF sampling and sensing technologies remains essential. MNs provide a promising strategy because they can selectively access the epidermal or dermal interstitial space while minimizing pain and tissue damage. Integration of MNs with biosensing platforms may therefore enable efficient ISF access and real-time glucose detection, providing a promising foundation for next-generation minimally invasive glucose monitoring systems [19].

3. Fundamentals of MNs Technology

At present, MNs are gaining increasing interest as a promising alternative to conventional blood-based diagnostics due to their micron-scale dimensions, minimal invasiveness, and reduced pain compared with standard hypodermic needles. In contrast to hypodermic needles, MNs penetration is typically limited to the superficial skin layers and does not reach dermal nerves or blood vessels, enabling painless and blood-free ISF collection, with sufficient ISF volumes for analysis obtainable within ∼1–10 min [39,40,41]. The interstitial fluid of the skin is found under the stratum corneum and is plentiful within the dermal layer, which has a thickness ranging approximately 1500–3000 µm [42]. MNs typically range from 25 to 2500 µm in heights that can penetrate the epidermis. They have a thickness of 50–200 µm for the extraction of ISF from the dermis and a tip diameter between 1 and 100 µm [43,44]. Importantly, their length should be appropriately designed to penetrate the stratum corneum (10–20 µm) [45] and reach the epidermal (50–150 µm) or dermal layers (500–2000 µm) [46]. Insertion pain can be minimized by optimizing MNs design parameters, including diameter, length, width, tip size, tip angle, and needle density. MNs with heights of approximately 500–900 µm are commonly used to reach the dermis for interstitial fluid extraction, providing safe and effective skin penetration depending on factors such as skin type, ethnicity, and age [47]. In addition, MNs length has the greatest influence on perceived pain, with longer MNs (>1450 µm) producing several-fold higher pain scores compared with shorter MNs (<700 µm) [47]. Currently, MNs are commonly fabricated from materials such as silicon, polymers, ceramics, and metals [48]. Recent advancements in micro/nano-fabrication, materials science, digital innovations, biosensing platforms, and flexible electronics have accelerated the development of integrated, miniaturized, and functional micro-sensing systems, particularly for MNs-based transdermal sensors used in point-of-care testing.

4. Types of MNs Structures and Mediated Interstitial Fluid Sampling

There are various types of MNs addressed in this review. MNs can be classified according to their design, fabrication strategies, and function, including applications in drug administration and biofluid extraction. They are commonly divided into six categories: solid, coated, hollow, porous, dissolving, and hydrogel-forming MNs (Figure 2). Materials used in MNs fabrication usually include silicon, stainless steel, biomaterials, and polymers [49,50]. The selection of an appropriate fabrication method depends on the chosen material, critical geometric parameters, and the intended functionality of the MNs. Another important consideration is the skin structure. Human skin is composed of three main layers: the epidermis, dermis, and hypodermis. Among these, the dermis contains a higher concentration of ISF, making it an optimal target for MNs-based biosensors for glucose monitoring. The comparative performance of representative MN-based optical glucose biosensors is reported in Table 2.

Figure 2. Schematic representation of solid, coated, hollow, porous, dissolving, hydrogel-forming MNs platforms for ISF access and integration with optical glucose biosensing systems.

4.1. Solid MNs

The solid MNs system was the first developed and is primarily composed of metals (including nickel, palladium, stainless steel, gold, and titanium), silicon, and various polymers [51]. They are easy to make and generally have great mechanical strength. Solid MNs pierce the skin to form microincisions, allowing body fluids to be collected. They are mechanically stable and easy to modify [52]. Solid MNs cannot penetrate deep enough to trigger pain receptors. As a result, its use is usually painless and may increase patient compliance [53]. To date, colorimetric or fluorescent solid MNs for glucose detection in ISF have not been reported, while existing approaches predominantly rely on electrochemical sensor integration. Previously, solid MNs fabricated from silicon and integrated with an electrochemical biosensor for glucose detection have been reported, demonstrating a detection range of 2–13.5 mM [54]. However, solid MNs have limitations in drug administration and ISF extraction due to their low drug-loading capacity and low fluid-collection effectiveness. To overcome these issues, other type of MNs such as hollow, porous, coated, dissolving, and hydrogel MNs were developed.

4.2. Coated MNs

Coated MNs are generally fabricated by depositing functional sensing materials onto mechanically robust solid MNs substrates, commonly composed of stainless steel, titanium, silicon, or polymers. The MNs surface can be modified with metals, conductive polymers, enzymes, hydrogels, or nanomaterials using dip-coating, spray-coating, electrodeposition, or layer-by-layer techniques [14]. This configuration combines the high mechanical strength of solid MNs with a functional coating that directly contacts ISF, allowing rapid analyte diffusion and signal generation. For glucose monitoring, coated MNs have predominantly been integrated with electrochemical sensing systems. Previously, gold-coated MNs functionalized with glucose oxidase (GOx) through a self-assembled monolayer demonstrated glucose detection over 30–400 mg/dL (approximately 1.7–22.2 mM) [55]. Non-enzymatic glucose sensing has also been demonstrated using Nafion and platinum-black-coated MNs electrodes, providing a linear detection range of 1–40 mM [56]. More recently, stainless-steel MNs coated with a GOx–chitosan layer and an alginate–polyvinyl alcohol hydrogel achieved glucose detection over 1–40 mM in artificial ISF [57]. However, coated MNs remain limited by non-uniform coating, coating delamination during skin insertion, restricted loading capacity, enzyme instability, and biofouling during prolonged use. These limitations have encouraged the development of alternative MNs platforms, including porous, hollow, dissolving, and hydrogel-forming MNs.

4.3. Hollow MNs

Hollow MNs can be fabricated from diverse materials, including metals (such as stainless steel, nickel, and titanium), polymers (e.g., polydimethylsiloxane and poly(methyl methacrylate)), silicon, glass, and ceramic materials (including silicon carbide and alumina) [58]. These hollow MNs feature microfluidic channels similar to those seen in ordinary hypodermic needles, allowing for the delivery of higher amounts of fluid. Previously, a hollow MNs array with integrated fluidic microchannels was fabricated on a bulk silicon wafer for in vivo ISF sampling and in situ colorimetric glucose detection. Mukerjee et al. designed hollow MNs with a 10 μm hole diameter and a shank height of 250–350 μm, which were successfully demonstrated for glucose detection by extracting ISF from the human earlobe via capillary action, with a sampling time of 15–20 min [16]. In another study, a vacuum-pressure-assisted blood-extraction system incorporating nickel hollow MNs was developed for in vivo glucose and cholesterol diagnosis. The device integrated blood collection, serum separation, and colorimetric paper-based sensing into a single platform, with all processes sequentially automated through one-touch activation [59]. Similarly, a hollow MNs integrated with a paper-based sensor was developed for colorimetric glucose detection. The MNs featured a sharp beveled edge and a length of 400 µm, enabling rapid fluid uptake via capillary action within 5 s. The device allowed simple and rapid glucose detection within 30 s and enabled discrimination between normal glycaemia (4–7 mM) and hyperglycaemia (>7 mM) [60].

4.4. Porous MNs

Porous MNs are designed with micro- or nanoscale pores either on their surface or throughout the entire needle body. Following skin insertion, these micro- or nanoscale pores can extract ISF through capillary action. The collected ISF can then be analyzed directly at the site of collection using surface functionalization or integrated sensing platforms [61,62]. Porous MNs are fabricated from biocompatible metals, ceramics, or biodegradable polymers such as polyethylene glycol (PEG) and poly(lactic-co-glycolic acid) (PLGA). Consequently, porous MNs are less prone to breakage than solid MNs. Several studies have reported porous colorimetric MNs integrated with optical sensing elements for glucose detection in ISF. For example, Zeng et al. developed porous PGMA-based colorimetric MNs functionalized with GOx and horseradish peroxidase (HRP), along with a 3,3′,5,5′-tetramethylbenzidine (TMB)-containing backing layer, allowing in situ visual glucose detection in rat skin with quantitative analysis over a 50–400 mg/dL range using RGB image analysis [63]. Interestingly, the porous MNs integrated with SERS probes and GOx was fabricated using PEG as a porogen and a monomer matrix, enabling in vivo glucose detection over a concentration range of 0–15 mM [64]. Furthermore, Zhu et al. extended this platform by developing a porous MNs integrated with SERS probes for multiplex detection of glucose, uric acid, and cholesterol, demonstrating detection ranges of 0–15 mM for glucose, 0–0.9 mM for uric acid, and 0–9 mM for cholesterol in ISF [65].

4.5. Dissolving MNs

Dissolving MNs are commonly referred to as “poke-and-release” systems. Biodegradable, bioresorbable, or water-soluble polymers such as polyvinyl alcohol (PVA), polyvinylpyrrolidone (PVP), cellulose derivatives, and sugars are often employed in the fabrication process, which is normally done via micro-molding technologies [66]. After MNs implantation into the skin, drug release begins through polymer dissolution, followed by gradual degradation of the biodegradable matrix [51]. The insertion of dissolving MNs forms microchannels in the skin that allow ISF to be extracted using a vacuum pump or captured by an absorbent membrane. Ito et al. developed dissolving MNs made from sodium chondroitin sulfate and dextran to collect ISF from rat skin for glucose monitoring using a moist nonwoven cloth, whereby increasing the cloth-skin contact time from 0.1 to 5.0 min significantly increased ISF glucose extraction from 1.99 to 5.04 µg, and normal skin physiology recovered within 1 h after MNs removal. This study also reported a strong correlation between the measured ISF glucose levels and blood-glucose concentrations [67]. Dissolving MNs help reduce biohazardous sharp waste and improve dosing accuracy but may exhibit inconsistent skin penetration and an increased risk of breakage during use [68,69].

4.6. Hydrogel-Forming MNs

Hydrogel-forming MNs are generally known as swelling MNs. They are fabricated by crosslinking two or more polymers, including chitosan, PVA, PLGA, hyaluronic acid (HA), methacrylated hyaluronic acid (MeHA), polymethyl vinyl ether-alt-maleic anhydride (PMVE/MA), and gelatin methacryloyl (GelMA), which are commonly preferred in fabrication methods [70]. These polymers can be cross-linked to form hydrogel MNs with more effective swelling, depending on the applied chemical, physical, or photo-crosslinking method. The polymer swells after skin application due to uptake of ISF [71]. The development of hydrogel MNs is based on the use of swelling or super-swelling polymers, which are often crosslinked to improve their ability to absorb fluids. There are several reports where the hydrogel-based MNs for glucose detection utilized the chosen different polymer material. For example, hydrogel MNs fabricated from MeHA and subsequently crosslinked via UV irradiation were shown to rapidly collect interstitial fluid. These MNs were able to collect 1.4 μL of ISF within 1 min and 2.3 μL within 10 min from mouse skin [39]. It has been reported that GelMA-based MNs showed swelling ratios of 293–423% and a compressive modulus of 3.34–7.23 MPa demonstrated the capability to extract glucose and vancomycin from ISF [40]. In addition, Bao et al. also utilized GelMA mixed with PVA at a 1:1 ratio to fabricate hydrogel MNs exhibiting a swelling rate exceeding 300% within 60 min. These hydrogel MNs enabled colorimetric detection of four biomarkers including pH, glucose, uric acid, and insulin and detected glucose over a concentration range of 0 to 24 mM in vitro. In vivo studies also showed that the MNs absorbed 45 mg of ISF within 10 min after application, resulting in visible color changes corresponding to the four biomarkers. Similarly, He et al. reported the development of a hydrogel MNs patch composed of PVA and chitosan for extracting multiple biomarkers, including glucose, chloride, lactate, and bovine serum albumin (BSA). The PVA and chitosan formulation at a 1:1 ratio exhibited the highest swelling capacity and extraction efficiency. For glucose detection, the MNs extracted 1.25 mg within 10 min in rabbit skin, and the micropores created in the skin disappeared rapidly within 3 min [72]. However, one limitation associated with hydrogel MNs is the difficulty in preserving the geometric integrity of the fabricated structures [73]. Furthermore, certain polymer systems require cross linking to enhance mechanical strength, regulate swelling behavior, and improve the overall stability of the MNs.

Table 2. Comparative performance of representative MN-based optical glucose biosensors.

MNs Type	Sensing Mechanism	Main Material/Sensing Element	Glucose Detection Range	Sensitivity/LOD	Response Time/Stability	Validation	Ref.
Solid MNs	MN-created microchannels + ISF extraction	Glass solid MNs	N.R.	N.R.	ISF extraction 2–10 min	Rat and human ISF	[68]
	Reverse iontophoresis + electrochemical sensing	Solid MN array + electrochemical sensor	N.R.	N.R.	Real-time	In vitro/in vivo	[74]
	Impedimetric glucose sensing	Polymeric MN + immobilized GOx	N.R.	N.R.	Up to ~7-day operation	ISF/tissue model	[70]
	Differential electrochemical sensing	Individually addressable MN electrodes	Physiological glucose range	N.R.	Real-time	Human ISF	[71]
	Enzymatic electrochemical	Polymeric MN + PtNP/PEDOT-PSS	2–20 mM	Enhanced current response	Real-time	In vivo; Clarke error grid	[72]
	Enzymatic amperometric	Composite MN + graphene/TTF/GOx	2–15 mM	Reported experimentally	Extended-use monitoring	Artificial skin, plasma, mice	[73]
Coated MNs	GOx amperometric sensing	Au-coated MN + GOx	30–400 mg/dL (~1.7–22.2 mM)	N.R.	Real-time	Transdermal prototype	[55]
	Non-enzymatic amperometry	Pt-black/Nafion-coated MN	1–40 mM	175 µA mM−1 cm−2; LOD 23 µM	~2 s	In vitro	[56]
	SERS	Ag-coated PMMA MN + 1-DT	0–20 mM	N.R.	Few minutes	Diabetic mice; 93% Clarke A + B	[75]
	Photonic structural color	FPBA/SiO2 colloidal-crystal coating	5.6–22.3 mM	N.R.	~5 min; reversible	Diabetic mice	[76]
	Enzymatic electrochemical CGM	Functional MN electrode + gel electrolyte	Physiological range	N.R.	Continuous monitoring	In vivo	[77]
	GOx amperometric	Stainless-steel MN + GOx/chitosan + Alg–PVA coating	1–40 mM	LOD 0.20 mM	Long-term stability reported	Artificial ISF/skin phantom	[57]
Hollow MNs	Capillary ISF extraction + glucose analysis	Silicon hollow MN	N.R.	N.R.	15–20 min	Human earlobe	[16]
	GOx/HRP colorimetry	Hollow MN + paper-based sensor	~0–10 mM	N.R.	~30-s detection	In vitro	[60]
	SERS	Ag-coated hollow agarose MN + 1-DT	5–150 mM	RMSE 5.1 mM	<10 s	Skin phantom	[78]
	Nanozyme electrochemical sensing	3D-printed hollow MN + nanozyme electrode	0.1 µM–50 mM	LOD 0.285 µM	Dynamic/wireless	Artificial ISF	[79]
Porous MNs	FAD-GDH amperometric	Porous Au MN + FAD-GDH	0.1–10 mM	50.86 µA cm−2 mM−1	~3 s; ~20% signal loss at 30 d	Artificial ISF	[61]
	GOx/HRP/TMB colorimetry	Porous PGMA MN	50–400 mg/dL (~2.8–22.2 mM)	N.R.	~10 min	Rat skin; RGB analysis	[63]
	Fluorescence, PBA recognition	Porous MN + fluorescent nanodiamond boronic hydrogel	0–36.1 mM	N.R.	Lag 3.8 ± 1.3 min ↑; 6.3 ± 3.3 min ↓	Mice and pigs	[80]
	GOx/HRP colorimetry	Porous PLA microsphere MN + paper sensor	1–20 mM	LOD 0.41 mM	Extraction 0.95 µL min−1 needle−1	Skin model	[81]
	SERS	Porous MN + plasmonic probe + GOx/4-MPBA	0–15 mM	Ratio-based SERS quantification	Stable/accurate detection	In vitro/in vivo	[64]
	Multiplex SERS	Porous MN + GNSs + 4-MPBA + GOx/UOx/ChOx	Glucose 0–15 mM	LOD 0.290 mM	~40 min	Rats	[65]
Dissolving/biodegradable MNs	Dissolution-assisted ISF collection + glucose assay	Chondroitin sulfate/dextran	N.R.	N.R.	0.1–5 min sampling	Rat skin	[67]
	Colorimetric dermal tattoo	HA dissolving MN + GOx/color reagents	~1–10 mM	Semiquantitative	~10 min	Rabbit	[82]
	Fluorescence; diboronic-acid recognition	Silk fibroin + fluorescent GF-monomer + PVA	2.8–25 mM	N.R.	15–20 min	Mice; smartphone readout	[83]
Hydrogel-forming/swelling MNs	ISF absorption + offline analysis	Crosslinked MeHA	Sampling platform	—	1.4 µL/1 min; 2.3 µL/10 min	Mouse skin	[39]
	ISF extraction + glucose analysis	PVA/chitosan hydrogel	N.R.	N.R.	~10 min	Rabbit skin	[84]
	Fluorescent aptamer	MeHA hydrogel + fluorescent aptamer	0–30 mM	LOD 1.1 mM	2 min; stable up to 14 d	Diabetic rats	[85]
	Photonic crystal/PBA recognition	PBA-functionalized photonic hydrogel	2–8 mM	LOD 1.07 mM	~30 min; reversible	Porcine skin	[86]
	Nanozyme-assisted colorimetry	Swelling hydrogel + AuNP/GOx/HRP/TMB	1–20 mM	N.R.	~10 min	Rats	[87]
	Multiplex enzymatic colorimetry	GelMA/PVA hydrogel	0–24 mM	N.R.	~10 min	T2D rats	[88]
	GOx/HRP/TMB colorimetry	PVA/CMCS hydrogel	Physiological/hyperglycemic range	N.R.	Hyperglycemia detection ~5 min	Rats	[89]
	FRET fluorescence	Glucose-responsive hydrogel	Physiological range	Sensitivity 0.029 mM−1; LOD 0.193 mM	~7.7 min	Serum/porcine skin	[90]
	MOF nanozyme colorimetry	Zr-MOF@Pt/TMB/GOx hydrogel MN	0.5–800 µM	LOD 0.45 µM	N.R.	In vitro/serum	[91]
	ISF extraction + electrochemical detection	Swellable hydrogel MN + external electrode	Clinically relevant range	N.R.	~7.06 µL ISF/5 min	In vitro/ex vivo	[92]
	GOx/PB electrochemical sensing	Swellable MN + PB/chitosan-SWCNT/GOx/Nafion	Clinically relevant range	12.26 µA mM−1 cm−2; LOD 0.08 mM	~5-min sampling	In vitro/ex vivo	[93]
	Dual-enzyme colorimetry	Hydrogel MN + GOx/HRP-loaded MAF-7	1–25 mM	N.R.	N.R.	Glucose detection model	[94]

Abbreviations: GOx, glucose oxidase; HRP, horseradish peroxidase; TMB, 3,3′,5,5′-tetramethylbenzidine; PBA, phenylboronic acid; PGMA, poly(glycidyl methacrylate); PLA, polylactic acid; MeHA, methacrylated hyaluronic acid; GelMA, gelatin methacryloyl; PVA, polyvinyl alcohol; PMMA, poly(methyl methacrylate); SERS, surface-enhanced Raman scattering; LOD, limit of detection; RMSE, root-mean-square error; N.R., not reported.

Table 2. Comparative performance of representative MN-based optical glucose biosensors.

MNs Type	Sensing Mechanism	Main Material/Sensing Element	Glucose Detection Range	Sensitivity/LOD	Response Time/Stability	Validation	Ref.
Solid MNs	MN-created microchannels + ISF extraction	Glass solid MNs	N.R.	N.R.	ISF extraction 2–10 min	Rat and human ISF	[68]
	Reverse iontophoresis + electrochemical sensing	Solid MN array + electrochemical sensor	N.R.	N.R.	Real-time	In vitro/in vivo	[74]
	Impedimetric glucose sensing	Polymeric MN + immobilized GOx	N.R.	N.R.	Up to ~7-day operation	ISF/tissue model	[70]
	Differential electrochemical sensing	Individually addressable MN electrodes	Physiological glucose range	N.R.	Real-time	Human ISF	[71]
	Enzymatic electrochemical	Polymeric MN + PtNP/PEDOT-PSS	2–20 mM	Enhanced current response	Real-time	In vivo; Clarke error grid	[72]
	Enzymatic amperometric	Composite MN + graphene/TTF/GOx	2–15 mM	Reported experimentally	Extended-use monitoring	Artificial skin, plasma, mice	[73]
Coated MNs	GOx amperometric sensing	Au-coated MN + GOx	30–400 mg/dL (~1.7–22.2 mM)	N.R.	Real-time	Transdermal prototype	[55]
	Non-enzymatic amperometry	Pt-black/Nafion-coated MN	1–40 mM	175 µA mM−1 cm−2; LOD 23 µM	~2 s	In vitro	[56]
	SERS	Ag-coated PMMA MN + 1-DT	0–20 mM	N.R.	Few minutes	Diabetic mice; 93% Clarke A + B	[75]
	Photonic structural color	FPBA/SiO2 colloidal-crystal coating	5.6–22.3 mM	N.R.	~5 min; reversible	Diabetic mice	[76]
	Enzymatic electrochemical CGM	Functional MN electrode + gel electrolyte	Physiological range	N.R.	Continuous monitoring	In vivo	[77]
	GOx amperometric	Stainless-steel MN + GOx/chitosan + Alg–PVA coating	1–40 mM	LOD 0.20 mM	Long-term stability reported	Artificial ISF/skin phantom	[57]
Hollow MNs	Capillary ISF extraction + glucose analysis	Silicon hollow MN	N.R.	N.R.	15–20 min	Human earlobe	[16]
	GOx/HRP colorimetry	Hollow MN + paper-based sensor	~0–10 mM	N.R.	~30-s detection	In vitro	[60]
	SERS	Ag-coated hollow agarose MN + 1-DT	5–150 mM	RMSE 5.1 mM	<10 s	Skin phantom	[78]
	Nanozyme electrochemical sensing	3D-printed hollow MN + nanozyme electrode	0.1 µM–50 mM	LOD 0.285 µM	Dynamic/wireless	Artificial ISF	[79]
Porous MNs	FAD-GDH amperometric	Porous Au MN + FAD-GDH	0.1–10 mM	50.86 µA cm−2 mM−1	~3 s; ~20% signal loss at 30 d	Artificial ISF	[61]
	GOx/HRP/TMB colorimetry	Porous PGMA MN	50–400 mg/dL (~2.8–22.2 mM)	N.R.	~10 min	Rat skin; RGB analysis	[63]
	Fluorescence, PBA recognition	Porous MN + fluorescent nanodiamond boronic hydrogel	0–36.1 mM	N.R.	Lag 3.8 ± 1.3 min ↑; 6.3 ± 3.3 min ↓	Mice and pigs	[80]
	GOx/HRP colorimetry	Porous PLA microsphere MN + paper sensor	1–20 mM	LOD 0.41 mM	Extraction 0.95 µL min−1 needle−1	Skin model	[81]
	SERS	Porous MN + plasmonic probe + GOx/4-MPBA	0–15 mM	Ratio-based SERS quantification	Stable/accurate detection	In vitro/in vivo	[64]
	Multiplex SERS	Porous MN + GNSs + 4-MPBA + GOx/UOx/ChOx	Glucose 0–15 mM	LOD 0.290 mM	~40 min	Rats	[65]
Dissolving/biodegradable MNs	Dissolution-assisted ISF collection + glucose assay	Chondroitin sulfate/dextran	N.R.	N.R.	0.1–5 min sampling	Rat skin	[67]
	Colorimetric dermal tattoo	HA dissolving MN + GOx/color reagents	~1–10 mM	Semiquantitative	~10 min	Rabbit	[82]
	Fluorescence; diboronic-acid recognition	Silk fibroin + fluorescent GF-monomer + PVA	2.8–25 mM	N.R.	15–20 min	Mice; smartphone readout	[83]
Hydrogel-forming/swelling MNs	ISF absorption + offline analysis	Crosslinked MeHA	Sampling platform	—	1.4 µL/1 min; 2.3 µL/10 min	Mouse skin	[39]
	ISF extraction + glucose analysis	PVA/chitosan hydrogel	N.R.	N.R.	~10 min	Rabbit skin	[84]
	Fluorescent aptamer	MeHA hydrogel + fluorescent aptamer	0–30 mM	LOD 1.1 mM	2 min; stable up to 14 d	Diabetic rats	[85]
	Photonic crystal/PBA recognition	PBA-functionalized photonic hydrogel	2–8 mM	LOD 1.07 mM	~30 min; reversible	Porcine skin	[86]
	Nanozyme-assisted colorimetry	Swelling hydrogel + AuNP/GOx/HRP/TMB	1–20 mM	N.R.	~10 min	Rats	[87]
	Multiplex enzymatic colorimetry	GelMA/PVA hydrogel	0–24 mM	N.R.	~10 min	T2D rats	[88]
	GOx/HRP/TMB colorimetry	PVA/CMCS hydrogel	Physiological/hyperglycemic range	N.R.	Hyperglycemia detection ~5 min	Rats	[89]
	FRET fluorescence	Glucose-responsive hydrogel	Physiological range	Sensitivity 0.029 mM−1; LOD 0.193 mM	~7.7 min	Serum/porcine skin	[90]
	MOF nanozyme colorimetry	Zr-MOF@Pt/TMB/GOx hydrogel MN	0.5–800 µM	LOD 0.45 µM	N.R.	In vitro/serum	[91]
	ISF extraction + electrochemical detection	Swellable hydrogel MN + external electrode	Clinically relevant range	N.R.	~7.06 µL ISF/5 min	In vitro/ex vivo	[92]
	GOx/PB electrochemical sensing	Swellable MN + PB/chitosan-SWCNT/GOx/Nafion	Clinically relevant range	12.26 µA mM−1 cm−2; LOD 0.08 mM	~5-min sampling	In vitro/ex vivo	[93]
	Dual-enzyme colorimetry	Hydrogel MN + GOx/HRP-loaded MAF-7	1–25 mM	N.R.	N.R.	Glucose detection model	[94]

Abbreviations: GOx, glucose oxidase; HRP, horseradish peroxidase; TMB, 3,3′,5,5′-tetramethylbenzidine; PBA, phenylboronic acid; PGMA, poly(glycidyl methacrylate); PLA, polylactic acid; MeHA, methacrylated hyaluronic acid; GelMA, gelatin methacryloyl; PVA, polyvinyl alcohol; PMMA, poly(methyl methacrylate); SERS, surface-enhanced Raman scattering; LOD, limit of detection; RMSE, root-mean-square error; N.R., not reported.

In addition to passive capillary and swelling mechanisms, external forces have been introduced to accelerate MNs-mediated ISF extraction. These approaches include vacuum pressure, osmotic pressure, electroosmosis, mechanical vibration, and integrated microfluidic pumps. Such strategies can increase sampling volume and shorten extraction time, which is particularly important because only microliter-scale volumes of ISF are typically available. Nevertheless, increasing extraction force may alter local tissue perfusion or introduce blood contamination, while prolonged extraction can affect the representativeness of the collected sample. Therefore, an optimal MNs sampling platform must balance extraction efficiency with tissue integrity, analyte recovery, and sampling reproducibility [14,95].

For glucose monitoring, the integration of ISF sampling and sensing within a single MNs platform is particularly attractive because it minimizes sample handling and may enable near-real-time or continuous measurements. Hollow and porous MNs can directly transport glucose-containing ISF toward electrochemical or optical sensing regions, whereas hydrogel-forming MNs can absorb glucose into a hydrated sensing matrix. Current challenges include obtaining sufficient and reproducible ISF volumes, maintaining unobstructed mass transport, preventing blood contamination and biofouling, minimizing sampling delay, and achieving stable calibration during prolonged skin contact. Addressing these limitations will be essential for translating MNs-mediated ISF sampling from laboratory prototypes into reliable wearable glucose-monitoring systems [14,18].

5. Fabrication Technologies for MNs Biosensors

Fabrication technology is a critical determinant of the geometry, mechanical strength, reproducibility, surface properties, and sensing performance of MNs-based optical biosensors. The selection of an appropriate manufacturing method depends on the MNs type, constituent material, target insertion depth, sensing mechanism, and required integration with electrodes, enzymes, nanomaterials, or optical reporters. Conventional fabrication approaches include micromolding, photolithography, etching, micromachining, and laser-based processing, whereas additive manufacturing and advanced surface-modification techniques have recently expanded the design flexibility of MNs biosensors [50]. The fabrication technologies for MNs biosensors were illustrated in Figure 3.

5.1. Micromolding

Micromolding is one of the most widely used methods for fabricating dissolving, and hydrogel-forming MNs. In this approach, polymer solutions or prepolymer mixtures are filled into negative molds, commonly prepared from polydimethylsiloxane (PDMS), followed by drying, polymerization, or crosslinking. Vacuum, centrifugation, or pressure-assisted filling may be employed to ensure complete filling of the needle cavities. Micromolding offers relatively low cost, simple processing, and good scalability; however, the final MN geometry is dependent on mold quality, and incorporation of sensing materials may require additional coating or immobilization steps [50].

5.2. Photolithography and Etching Techniques

Photolithography and etching techniques are frequently employed for the fabrication of silicon and metallic MNs. Photolithography enables precise definition of MNs dimensions and array patterns, while wet or dry etching can generate sharp and highly reproducible structures. Deep reactive ion etching is particularly suitable for producing high-aspect-ratio silicon MNs with controlled geometry. These methods provide excellent dimensional accuracy and are compatible with microelectronic fabrication, making them attractive for electrochemical biosensors. Nevertheless, they generally require cleanroom facilities, specialized equipment, multiple processing steps, and relatively high manufacturing costs [50].

5.3. Laser-Based Fabrication and Micromachining

Laser-based fabrication and micromachining provide alternative routes for manufacturing metallic and polymeric MNs. Laser cutting, laser ablation, and laser micromachining can rapidly generate MNs structures from stainless steel, titanium, polymers, or composite materials without the need for complex masks. These approaches offer flexibility in modifying needle length, width, spacing, and tip geometry. However, thermal damage, surface roughness, and limited resolution may occur depending on the laser parameters and material properties [50].

5.4. Droplet-Born Air Blowing

Droplet-born air blowing (DAB) is a mold-free fabrication technique developed for dissolving MNs. In this approach, controlled droplets of a water-soluble polymer are sequentially dispensed onto a substrate, brought into contact with an opposing surface, and elongated into MNs structures. Air blowing is subsequently applied to dry and solidify the elongated droplets while maintaining the needle geometry. The droplet volume can be precisely controlled by adjusting the dispensing pressure and time, allowing reproducible control of MNs dimensions and incorporated material content. Importantly, DAB operates under mild conditions of approximately 4–25 °C and enables rapid MNs fabrication within 10 min, without requiring heating or ultraviolet irradiation. These characteristics are advantageous for incorporating temperature- or light-sensitive biomolecules, including enzymes and other biological sensing components. However, precise control of droplet rheology, polymer viscosity, drying conditions, and needle geometry is required to obtain reproducible MNs arrays [96].

5.5. Three-Dimensional (3D) Printing

3D printing has emerged as a promising technology for the rapid and customizable fabrication of MNs biosensors. Techniques such as stereolithography, digital light processing, two-photon polymerization, and fused deposition modeling enable the production of complex MNs geometries directly from computer-aided designs (CADs). In particular, high-resolution photopolymerization techniques allow precise control over needle height, base diameter, tip radius, and array configuration. 3D printing also facilitates rapid prototyping and patient-specific design. However, limitations remain regarding printing resolution, biocompatibility of photopolymerizable resins, post-processing requirements, and the electrical conductivity of printed materials [97].

For biosensing applications, MNs fabrication is commonly followed by surface functionalization or coating to introduce sensing components. Conductive metals, carbon-based materials, conducting polymers, enzymes, antibodies, aptamers, fluorophores, and nanomaterials can be deposited using dip coating, spray coating, electrodeposition, sputtering, inkjet printing, or layer-by-layer assembly. For glucose sensing, GOx or glucose dehydrogenase (GDH) is frequently immobilized on conductive MNs electrodes, while optical platforms may incorporate fluorescent probes, plasmonic nanoparticles, or colorimetric reagents. The uniformity and stability of these functional layers are essential because coating defects or delamination during skin insertion can substantially reduce analytical sensitivity and reproducibility [14,36]. Future manufacturing strategies are expected to combine high-resolution additive manufacturing with automated surface functionalization, conductive materials, and scalable roll-to-roll or batch-processing techniques. Such integrated approaches will be important for achieving reproducible, low-cost, and clinically translatable MNs biosensors for continuous glucose monitoring in interstitial fluid [14].

Figure 3. Schematic illustration of MNs fabrication techniques: (A) Micromolding [98], (B) photolithography and etching [99], (C) laser-based fabrication and micromachining [100], (D) droplet-born air blowing [96], and (E) three-dimensional (3D) printing [101]. Adapted from Refs. [96,98,99,100,101] under the Creative Commons Attribution (CC BY 4.0) license.

Figure 3. Schematic illustration of MNs fabrication techniques: (A) Micromolding [98], (B) photolithography and etching [99], (C) laser-based fabrication and micromachining [100], (D) droplet-born air blowing [96], and (E) three-dimensional (3D) printing [101]. Adapted from Refs. [96,98,99,100,101] under the Creative Commons Attribution (CC BY 4.0) license.

6. Principles of Glucose Biosensing

Glucose biosensors convert glucose concentration into a measurable analytical signal through the combination of a recognition element and a transducer. The recognition component selectively interacts with glucose, whereas the transducer converts this interaction into an electrical, optical, or other physicochemical signal. In MNs-based systems, glucose sensing is predominantly achieved through electrochemical or optical mechanisms following direct contact with ISF. According to the glucose-recognition mechanism, sensing platforms can be broadly classified as enzymatic and non-enzymatic systems [14,102]. The principles of glucose biosensing were illustrated in Figure 4.

Figure 4. Schematic illustration of the principles of glucose biosensing, including enzymatic, non-enzymatic, and glucose-selective recognition mechanisms.

6.1. Enzymatic Glucose Sensing

Enzymatic glucose biosensors use glucose-specific oxidoreductases to catalyze glucose conversion into products that can subsequently be quantified. GOx and GDH are the most widely employed enzymes because of their high catalytic activity and glucose-recognition capability. In electrochemical systems, glucose oxidation generates electron transfer that can be measured as current, whereas optical systems may detect changes in oxygen, hydrogen peroxide, pH, fluorescence, or color produced during the enzymatic reaction. Enzyme-based sensors generally provide high sensitivity and selectivity but are susceptible to enzyme denaturation, environmental pH and temperature, immobilization conditions, and long-term stability limitations [102,103]. Enzymatic electrochemical glucose biosensors are commonly described in three generations according to the electron-transfer mechanism. First-generation sensors use molecular oxygen as the natural electron acceptor and quantify oxygen consumption or hydrogen peroxide formation. Second-generation systems employ artificial redox mediators, such as ferrocene derivatives, ferricyanide, or osmium complexes, to facilitate electron transfer between the enzyme and electrode and reduce oxygen dependence. Third-generation sensors attempt direct electron transfer between the enzyme redox center and electrode surface without a diffusible mediator, potentially simplifying sensor architecture and lowering operating potential [14,103].

6.1.1. Glucose Oxidase-Based Sensing

GOx is the most extensively employed enzyme in glucose biosensors because of its high glucose selectivity, catalytic activity, relative stability, and established immobilization chemistry. GOx contains flavin adenine dinucleotide (FAD) as its redox cofactor and catalyzes the oxidation of β-D-glucose to glucono-δ-lactone while FAD is reduced to FADH₂. Molecular oxygen subsequently reoxidizes FADH₂, generating hydrogen peroxide, which can be electrochemically or optically quantified as an indirect measure of glucose concentration [14,102]. GOx is particularly attractive for MNs biosensors because it can be immobilized onto metallic electrodes, conductive polymers, nanomaterials, or hydrogel matrices positioned directly within the ISF-accessing region. However, first-generation GOx systems depend on local oxygen availability, which may vary within skin tissue and introduce measurement error. Hydrogen peroxide detection may also require relatively high electrode potentials, increasing interference from electroactive compounds such as uric acid and ascorbic acid. Mediators, nanostructured electrodes, and selective membranes are therefore frequently incorporated to improve electron transfer, decrease operating potential, and enhance glucose selectivity [14,95].

6.1.2. Glucose Dehydrogenase-Based Sensing

GDH provides an alternative enzymatic approach that is largely independent of dissolved oxygen. Different GDH families use cofactors including pyrroloquinoline quinone (PQQ), nicotinamide adenine dinucleotide (NAD), and flavin adenine dinucleotide (FAD). During glucose oxidation, these cofactors accept electrons that are subsequently transferred to an electrode through mediators or direct electron-transfer pathways. The oxygen independence of GDH can be advantageous for ISF sensing, where local oxygen concentrations may fluctuate [95,104]. Nevertheless, each GDH system presents specific limitations. PQQ-GDH exhibits rapid electron-transfer kinetics but may respond to other sugars, including maltose, galactose, and xylose, depending on the enzyme variant. NAD-dependent GDH requires an additional diffusible cofactor and efficient electrochemical regeneration of NADH, complicating sensor design. FAD-GDH provides improved substrate specificity and oxygen-independent activity and has therefore attracted increasing interest for glucose biosensors. Appropriate enzyme selection is particularly important for MNs-based continuous monitoring, where long-term selectivity, catalytic stability, and resistance to physiological interference are essential [105,106].

6.2. Non-Enzymatic Glucose Sensing

Non-enzymatic glucose sensing eliminates biological enzymes and typically relies on direct catalytic oxidation of glucose at an electrode surface. Noble and transition metals, including Pt, Au, Ni, and Cu, as well as their oxides, alloys, sulfides, carbon composites, metal-organic frameworks, and other nanostructured materials have been investigated. Their high surface area and catalytic activity can improve electron transfer and increase glucose sensitivity. Compared with enzymatic sensors, non-enzymatic systems generally offer greater thermal and chemical stability, simpler storage, and reduced susceptibility to enzyme degradation [103,104]. However, direct glucose oxidation frequently requires specific electrode potentials or alkaline conditions and may suffer from interference by electroactive compounds and other carbohydrates. Chloride adsorption and surface fouling may additionally reduce catalytic activity in physiological fluids. These issues are particularly important for MNs biosensors operating directly in ISF. Consequently, nanostructuring, alloy formation, protective membranes, and surface functionalization are increasingly employed to enhance catalytic activity and selectivity under physiologically relevant conditions [103].

6.3. Recognition Elements and Glucose-Selective Materials

Beyond enzyme-mediated catalysis, glucose recognition can be achieved using synthetic receptors and functional materials. Phenylboronic acid (PBA) and diboronic acid derivatives are particularly important for optical glucose sensing because boronic acids reversibly bind cis-diol groups of saccharides to form boronate esters. Binding can induce changes in fluorescence intensity, wavelength, color, swelling, or other physicochemical properties that are translated into glucose-dependent signals. Diboronic acid architectures have been developed to improve glucose selectivity by providing two appropriately positioned boronic acid recognition sites [107]. Other recognition strategies include molecularly imprinted polymers (MIPs), which contain synthetic binding cavities complementary to glucose, and catalytic nanomaterials that combine recognition and transduction functions. For optical MNs biosensors, glucose-responsive hydrogels, fluorophores, plasmonic nanoparticles, and Raman-active substrates may be incorporated directly into the MNs structure or sensing layer. Selection of the recognition material should consider glucose affinity within the physiological ISF range, reversibility, resistance to interference, biocompatibility, and stability during repeated or continuous monitoring [103,107].

7. Microneedle-Based Electrochemical Glucose Biosensors

MNs-based electrochemical biosensors combine minimally invasive access to ISF with electrochemical transduction, allowing biochemical reactions occurring at the MNs–ISF interface to be converted into measurable electrical signals. Compared with conventional blood-based electrochemical glucose testing, MNs platforms can provide direct access to dermal ISF while minimizing tissue injury and enabling repeated or continuous measurements. Electrochemical MNs sensors are particularly attractive because of their high sensitivity, rapid response, low power consumption, ease of miniaturization, and compatibility with wearable electronics. Depending on the transduction principle, these systems can operate through amperometric, potentiometric, or impedimetric detection, although amperometric sensing currently dominates MNs-based glucose monitoring [108]. The MNs-based electrochemical glucose biosensors were illustrated in Figure 5.

Figure 5. Schematic classification of MNs-based electrochemical glucose biosensors, including (A) amperometric, (B) potentiometric, and (C) impedimetric sensing modes for interstitial-fluid glucose detection.

7.1. Amperometric Sensing

Amperometric sensing is the most extensively investigated electrochemical approach for MNs-based glucose monitoring. In this method, a constant potential is applied between the working and reference electrodes, and the resulting current generated by an oxidation or reduction reaction is measured. For GOx-based sensors, glucose is enzymatically oxidized to gluconolactone, accompanied by the generation of hydrogen peroxide (H₂O₂) or reduction of an electron-transfer mediator. The electrochemical oxidation or reduction of these products generates a current proportional to glucose concentration. Because amperometric measurements provide high sensitivity, rapid response, and straightforward miniaturization, they are highly suitable for continuous ISF glucose monitoring [109].

7.2. Potentiometric Sensing

Potentiometric sensing measures the potential difference between an indicator electrode and a stable reference electrode under conditions of negligible current flow. The resulting potential is generally governed by the activity of a target ionic species according to the Nernst relationship. Potentiometric MNs sensors are therefore well established for monitoring electrolytes such as Na⁺, K⁺, and H⁺ in ISF. However, direct potentiometric glucose detection is less common because glucose is electrically neutral and does not itself generate an ion-selective potential. Consequently, potentiometric glucose biosensors typically require an indirect recognition mechanism in which enzymatic glucose conversion produces changes in proton concentration, ionic composition, or interfacial potential that can be measured potentiometrically [108,110].

7.3. Impedimetric Sensing

Impedimetric sensing measures changes in the electrical impedance of the electrode–solution interface over a defined frequency range, most commonly using electrochemical impedance spectroscopy (EIS). Unlike conventional amperometry, EIS can monitor changes in interfacial resistance, capacitance, dielectric properties, or charge-transfer processes using relatively small excitation signals. Glucose recognition may therefore be translated into concentration-dependent changes in impedance through enzyme immobilization, glucose-responsive polymers, molecularly imprinted layers, or changes occurring at the electrode interface [108].

8. Microneedle-Based Optical Glucose Biosensors

Recently, MNs-based sensors have been developed by integrating biosensing elements onto MNs surfaces or within their structures for real-time detection of biomarkers in ISF or blood. MNs and biosensor technologies have emerged as innovative platforms offering minimal pain, precise delivery, improved bioavailability, and increased user adherence. It can also detect a wide range of biomarkers, including electrolytes, nucleic acids, polysaccharides, proteins, and drugs, as well as metabolites including glucose in ISF [20,111].

8.1. Fundamentals of Optical Glucose Sensing

MNs-based sensing platforms integrate two fundamental functions: ISF extraction or access and biochemical glucose detection through an integrated sensing mechanism. For direct ISF sampling using MNs, the extraction process generally involves two primary stages: (1) successful insertion of the MNs through the stratum corneum into the viable epidermis or superficial dermis, and (2) subsequent accumulation or collection of ISF. ISF harvesting can be achieved through multiple strategies, including coupling the MNs array with an external absorbent reservoir (e.g., paper-based collectors) to wick fluid from the skin, or employing driving forces such as suction-assisted extraction, pressure-driven convection, passive diffusion, or osmotic gradients [112]. Glucose detection in MNs-based systems relies on specific molecular-recognition elements integrated with an appropriate signal-transduction mechanism. The detection principle varies according to the sensing modality and can be broadly categorized into enzymatic detection, non-enzymatic detection, and aptamer-based recognition strategies.

Optical glucose sensing relies on conversion of glucose recognition into a measurable change in an optical property, including fluorescence intensity or lifetime, visible color, Raman spectral intensity, reflectance wavelength, or refractive index (Figure 6). Glucose recognition can occur enzymatically, most commonly through GOx, or non-enzymatically using synthetic receptors such as PBA. In GOx-based systems, glucose oxidation produces gluconolactone and H₂O₂, and the resulting products can trigger fluorogenic, chromogenic, or Raman-active reactions. Alternatively, reversible binding between boronic acid groups and glucose cis-diols can alter hydrogel swelling, fluorophore behavior, or photonic structure, thereby generating an optical response without enzymatic catalysis [23]. In aptamer-integrated MNs platforms, single-stranded DNA or RNA aptamers are engineered to selectively recognize glucose. Nevertheless, as aptamers serve as molecular recognition elements and lack intrinsic signal-generating capability, they require integration with a suitable signal transducer to translate target binding into a quantifiable response [15].

Figure 6. Schematic illustration of sensor-integrated MNs platforms for optical glucose detection in interstitial fluid: (A) colorimetric MNs, (B) fluorometric MNs, and (C) surface-enhanced Raman scattering (SERS) MNs.

Developments of MNs sensing platforms are primarily integrated with electrochemical and optical detection methods [113,114]. However, electrochemical MNs sensors exhibit several limitations, including signal drift, electrode degradation, susceptibility to biofouling and electrochemical interference, frequent calibration requirements, and limited multiplexing capability when compared with optical MNs sensors [20,115,116]. Apart from the electrochemical MNs sensor, the optical MNs sensor exhibits strong resistance to electromagnetic interference and does not rely on a reference electrode. It prevents potential drift and provides a stable signal for long-term monitoring, as well as more detailed chemical insights. Additionally, optical analysis minimizes disturbance to the biological environment while avoiding stability issues correlated with electrolytic reactions or electrode passivation.

8.2. Optical MNs Sensor Platforms for Glucose Detection

Optical MNs-based sensors represent a promising approach for minimally invasive glucose monitoring in ISF because of their high sensitivity, versatility, and compatibility with non-electrical signal readout. According to the optical signal-transduction mechanism, these sensors are generally classified into three main categories, such as fluorescent, colorimetric, and surface-enhanced Raman scattering (SERS) sensors (Figure 7). In general, optical biosensors operate through chemical or biological interactions that induce measurable changes in optical properties, such as light absorbance or emission. MNs-based optical-sensors fabrication processes are comprised of four features: the target analyte, the probe-sensing element, signal acquisition, and data readout (Figure 7). Target analytes are selectively recognized by probes and converted into measurable signals through a transducer, which are then displayed on readable devices [22]. Optical biosensors have gained widespread attention due to their versatility, high sensitivity and specificity, affordability, and portability, as well as their compatibility with MNs platforms. Importantly, optical transducers offer the potential for label-free and real-time detection.

Figure 7. Schematic representation of the fabrication process of MN-based sensors, showing target analytes, probes, sensor types, and data collection and analysis [22].

The operating mechanisms of glucose biosensors are typically based on either oxygen consumption or hydrogen peroxide production resulting from glucose oxidation during the enzyme-catalyzed reaction, as shown in Equation (1) [117].

$$\begin{array}{c}\mathrm{G}\mathrm{l}\mathrm{u}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{e}+O_2+H_{2}O\stackrel{\mathrm{G}\mathrm{l}\mathrm{u}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{e} \mathrm{O}\mathrm{x}\mathrm{i}\mathrm{d}\mathrm{a}\mathrm{s}\mathrm{e}}{\to }\mathrm{G}\mathrm{l}\mathrm{u}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{i}\mathrm{c} \mathrm{A}\mathrm{c}\mathrm{i}\mathrm{d}+H_2{O}_2\end{array}$$

(1)

GOx serves as the enzymatic recognition element in MNs-based sensors. However, when applied in MNs sensors, the inherent instability of GOx is a significant challenge, as its enzymatic activity can be degraded during the coating steps on the MNs surface and upon exposure to biological fluids [118]. In contrast, non-enzymatic electrocatalysts, such as metal-based nanomaterials, have also been applied in MNs sensors [119]. The following sections provide a summary of glucose optical-sensing approaches used in MNs systems ISF, including colorimetric, fluorescence, and SERS methods (Table 3).

Table 3. Comparison of MNs-based optical sensing strategies for glucose monitoring in ISF.

Optical Readout	Detection Principle	Matrix Material	Sensing Mechanism	ISF Volume	Detection Time (min)	Detection Range (mM)	Test Subject	Ref.
Colorimetry	MNs + Off-line recycling detection	PVA + CS	Non-enzymatic	1.25 ± 0.37 mg	10	0.02–0.5	Rabbits	[84]
	MNs + Off-line recycling detection	MeHA	Non-enzymatic	2.3 ± 0.4 mg	-	1–16	Mice	[39]
	MNs + paper-based sensor	GelMA + PVA	GOx + HRP	45 mg	10	0–20	T2D rat model	[88]
	MNs + paper-based sensor	MeHA + HA	GOx + HRP	16.22 µL	10	0–16	Human	[113]
	Dermal tattoo	HA	GOx + HRP	-	10	0–10	Rabbits	[82]
	MNs + µPADs	-	GOx + HRP	-	40	0.1–6	Ex vivo (pig skin)	[120]
	MNs + Sensor	MeHA	GOx-like PVA-AuNPs + HRP	60.2 mg	10	1–20	Rats	[87]
	MNs + Sensor	GelMA	GOx + HRP	-	-	1.7–21	In vitro (agarose-based skin model)	[121]
	MNs + Sensor	PVA	GOx + HRP-CaP	-	10	3–22	Diabetic mice	[122]
	MNs + Sensor	GelMA	GOx-MnO2@GO with dual-enzyme activity	5.1 ± 0.3 mg	5	0–27.8	Rats	[123]
	MNs + sensor	PVA + CMCS	GOx + PVA-CS-HRP	11.1 mg	10	2–20	Rats	[89]
	MNs + paper-based sensor	UV-crosslinked methacrylate polymer (GMA/TMPTMA/TEGDMA)	GOx + HRP	-	0.5	0–10	In vitro (Parafilm membrane)	[60]
	MNs + paper-based sensor	PEG + GMA	GOx + HRP	34.57 ± 4.76 mg	10	1.39–22	Rats	[63]
	MNs + paper-based sensor	PEGDA multi-groove MNs	GOx + catalase	-	2	0–110	Mice
	MNs + color-development beads	PEGDA HMNs	GOx + HRP	-	-	0–12	Ex vivo (pig skin)	[124]
	MNs + sensor	Resin SMNs coated with colloidal crystals (SiO2 NPs)	Non-enzymatic (FPBA-based colloidal-crystal colorimetric system)	-	5	5.6–22.3 mM	Diabetic mice	[76]
	MNs + sensor	PEGDA and AM MNs with colloidal crystals (SiO2 NPs)	Non-enzymatic (PAM/FPBA/histamine aptamer-based colloidal crystals)	-	30	LOD: 8 for pH, 1 mg mL−1 for glucose, 2 µg mL−1 for histamine	Rats	[125]
Fluorescence	MNs + sensor	SS SMNs	Fluorescently labeled E. coli glucose-binding protein (GBP)	-	30 s	3–40 mM	Ex vivo (pig skin)	[27]
	MNs + sensor	MeHA HMNs Aptamer-bonded MeHA hydrogel	Fluorescently labeled aptamer probe	-	2	0–30 mM for glucose LOD: 1.1 mM for glucose, 0.1 mM for ATP, 3.5 µM for L-tyrosinamide, and 25 nM for thrombin	Diabetic rats	[85]
	MNs + sensor	Nanodiamond-based boronic HMN	PBA functionalized hydrogel with fluorescent nanodiamonds	-	-	0–36.1 mM	Mice	[80]
		Silk + GF-monomer MNs	Diboronic acid and anthracene functionalized hydrogel	-	15–20	2.8–25 mM	Mice	[83]
Raman/SERS		Ag-coated SS HMNs	Non-enzymatic	-	N.R.	5–150	In vitro (skin phantoms)	[126]
		Ag-coated agarose HMNs	Non-enzymatic (1-DT)	-	10 s	5–150	In vitro (skin phantoms)	[78]
		Ag-coated PMMA MNs	Non-enzymatic (1-DT)	-	15	0–20	Type I diabetic mice	[75]
	MNs + Sensor	Monomer + porogen + SERS probe MNs	GNSs + 4-MPBA + GOx/UOx/ChOx	14	40 min	Detection range: 0–20 mM for glucose, 0–1.2 mM for uric acid, and 0-mM for cholesterol. LOD: 0.290 mM for glucose, 0.107 mM for uric acid, 0.526 mM for cholesterol.	Rats	[65]

Abbreviation: LOD: limit of detection; SMNs: solid microneedles; HMNs: hollow microneedles; SS: stainless steel; NPs: nanoparticles; HA: Hyaluronic acid; GelMA: Methacrylate-modified gelatin; GOx-like PVA-AuNPs: Polyvinyl alcohol-modified gold nanoparticles with glucose-like oxidase activity; μPAD: microfluidic paper-based analytical device; GOx-MnO₂@GO: GOx-conjugated MnO₂/graphene oxide; CMCS: carboxymethyl chitosan; PMMA: poly(methyl methacrylate); 1-DT: 1-decanethiol; E. coli: Escherichia coli; GNSs: gold nanoshells; PAM: polyacrylamide; FPBA: fluorophenylboronic acid; GBP: glucose-binding protein; TMPTMA: trimethylolpropane trimethacrylate; PEGDA: poly(ethylene glycol) diacrylate; PEG: polyethylene glycol; GMA: Glycidyl methacrylate: GF-monomer: glucose-responsive fluorescent monomer; GOx: glucose oxidase; Uox: urate oxidase; ChOx: cholesterol oxidase.

8.3. Colorimetric MNs Sensors for Glucose Detection

Colorimetric signals can be analyzed by the naked eye or captured using a smartphone and quantitatively analyzed via RGB value. The integration of colorimetric strategies with MNs patches facilitates efficient sampling and detection of biomarkers from ISF. Color generation originates from analyte-induced changes in the chemical structure, density, morphology, size of chromogenic agents, or orientation. According to the underlying color-development mechanisms, chromogenic systems can be broadly categorized into redox-responsive agents and nanostructure-based platforms, which commonly rely on redox reactions, chromogenic substrates, or structural color changes [127]. Most colorimetric MNs patches are designed as redox-responsive systems that rely on enzymatic reactions and incorporate nanostructures such as gold nanoparticles or nanozymes to enhance catalytic efficiency and amplify colorimetric signals.

According to the color change induced by redox-responsive agents, colorimetric sensing relies on enzyme-mediated reactions that generate visually observable signals. Glucose is initially oxidized by glucose oxidase, producing gluconic acid and H₂O₂. The generated H₂O₂ subsequently reacts with H₂O₂-responsive chromogenic systems, such as iodide [128] or HRP coupled with TMB, resulting in color changes visible to the naked eye [129]. Meanwhile, gluconic acid directly causes a decrease in the local solution pH, which can be monitored using a pH-responsive biosensor. For example, Li et al. reported the first development of MNs patch based on AuNP nanozymes with GOx-like activity for in situ glucose detection in ISF [87]. The MNs consisted of two layers: a tip layer embedded with AuNPs and GOx, which generated H₂O₂ as a byproduct of glucose oxidation, and a backing layer encapsulating HRP, which catalyzed the oxidation of TMB by H₂O₂ to form oxTMB, resulting in a visible color change. This MNs patch can collect 60.2 mg of ISF in vitro within 10 min, while in vivo the MNs exhibited a blue-green color in the ISF of hyperglycemic rats and the glucose level by analyzing the RGB value (Figure 8A).

Figure 8. MNs arrays integrated with a colorimetric sensor platform for transdermal glucose monitoring. (A) Swellable colorimetric MNs for glucose detection based on GOx-like gold nanoparticles [87]. (B) Glucose detection: in situ colorimetric analysis with double-layer hydrogel MNs patch based on polyvinyl alcohol and carboxymethyl chitosan [89]. (C) Wearable MNs patch for colorimetric detection of multiple signature biomarkers in vivo toward diabetic diagnosis [88]. (D) Colloidal crystal MNs patch for glucose monitoring [76]. (E) Colorimetric dermal-tattoo biosensor fabricated by MNs patch for multiplexed detection of health-related biomarkers [82]. (F) Controllable-pore porous MNs for high-speed extraction and biomarker detection of ISF [81]. Adapted from Refs. [81,82] under the Creative Commons Attribution (CC BY 4.0) license.

In 2024, the same research group further reported a double-layer hydrogel MNs patch for in situ colorimetric detection of hyperglycemia [89]. The tip layer (PVA/CMCS) contained GOx, while the backing layer consisted of an immobilized poly(vinyl alcohol)–chitosan–horseradish peroxidase (PVA-CS-HRP) hydrogel loaded with TMB. The MNs tip array exhibited good swelling properties, enabling the extraction of 10 mg of ISF within 10 min. The MNs patch identified hyperglycemia in rats within 5 min after insertion, and the rat skin fully recovered within 10 min (Figure 8B).

For multibiomarker detection, Bao et al. demonstrated a colorimetric MNs patch based on a double-crosslinked hydrogel (GelMA mixed with PVA) for the detection of four type 2 diabetes (T2D)-related biomarkers, including glucose, uric acid, insulin, and pH [88]. The MNs absorbed approximately 45 mg of ISF within 10 min in vivo. After MNs application to diabetic rats, the glucose concentration was estimated to be 18.63 mM based on color changes, which was close to the glucose meter reading of 21.9 mM (Figure 8C). Additionally, Gu et al. developed rigid-resin MNs decorated with colloidal crystal structures for glucose visualization, in which periodic SiO₂ nanoparticle nanostructures were embedded within a glucose-responsive fluorophenylboronic acid (FPBA) matrix [76]. When glucose diffused into the FPBA-SiO₂ colloidal crystals, swelling of the FPBA matrix increased the spacing between the periodically arranged SiO₂ nanoparticles, resulting in a redshift of the MNs from blue (200 mg dL⁻¹ glucose) to green (400 mg dL⁻¹ glucose). No significant inflammation was observed after GCC–MNs insertion, demonstrating the good biosafety of the GCC–MNs patch (Figure 8D).

Furthermore, He et al. developed a multiplex MNs-based skin-tattoo sensor on rabbit skin by encapsulating colorimetric reagents within hyaluronic-acid MNs [82]. This platform enabled the simultaneous detection of multiple health-related biomarkers, including pH, glucose, uric acid, and body temperature. For glucose sensing, the biosensor exhibited a distinct colorimetric response, with the glucose-responsive region changing from yellow at 6 mM to red at 8 mM, while the color signals corresponding to the other biomarkers remained unchanged. However, this colorimetric MNs sensor achieved only semiquantitative detection because of its limited sensitivity and accuracy. In addition, the safety of dermal tattoo-based sensing remains controversial, as the potential leakage of dye molecules and H₂O₂ into the skin raises biocompatibility concerns (Figure 8E).

Interestingly, Qin et al. reported the fabrication of porous MNs composed of PLA microspheres that exhibited sufficient mechanical strength to penetrate human skin and achieved an in vitro ISF extraction rate of 0.95 μL min⁻¹ per needle, representing the highest value reported to date as of 2025 [81]. Furthermore, glucose detection was demonstrated using a porous MNs-integrated paper sensor with a diffusion layer applied to a skin model containing different glucose concentrations (1–20 mM), achieving an LOD of 0.41 mM. The colorimetric response of the system originated from the immobilization of GOx and HRP-TMB on filter paper. However, the flexibility of the porous MNs, particularly the substrate component, should be further improved to enhance conformal contact with skin and overall system performance (Figure 8F).

8.4. Fluorescent MNs Sensors for Glucose Detection

The integration of MNs with fluorescence biosensors represents a promising advancement in biomedical applications. Fluorescent biosensors are widely employed in biological research for chemical detection due to their rapid response, high sensitivity, and strong specificity. These sensors can incorporate biological fluorophores, synthetic dyes, or nanoparticles as fluorescent probes. Fluorescence-based MNs sensors offer enhanced sensitivity and precise target recognition, enabling accurate quantitative analysis while reducing background interference via the use of specifically designed probes. Various types of MNs have been combined with fluorescence-based sensors to improve detection sensitivity and selectivity. In 2022, Zheng et al. developed a methacrylated hyaluronic acid (MeHA) hydrogel microneedles (HMNs) biosensor incorporating fluorescently labeled aptamer probes as recognition elements for the on-needle, and reagent-free detection of biomarkers, including glucose, adenosine triphosphate (ATP), L-tyrosinamide, and thrombin [85]. The team employed a strand-displacement strategy in which fluorophore-conjugated aptamers were hybridized with complementary DNA competitor strands labeled with a quencher to enable reagentless sensing. These MNs sensor patches demonstrated rapid in vitro glucose detection within 2 min and maintained stability for up to 14 days when stored at room temperature. Furthermore, the system enabled accurate and sensitive monitoring of both rising and falling glucose levels in diabetic animal models, particularly in the severe hypoglycemic range (Figure 9A).

Figure 9. MNs arrays using fluorescence methods for transdermal glucose monitoring. (A) HMN-assisted assay integrating aptamer probes and fluorescence detection for reagentless biomarker quantification [85]. (B) Fluorescent-based biodegradable MNs-sensor array for tether-free continuous glucose monitoring with smartphone application [83]. Adapted from Ref. [83] under the Creative Commons Attribution (CC BY 4.0) license.

In 2023, Sang et al. presented a fluorescence-based biodegradable MNs sensor for glucose detection, fabricated from silk fibroin embedded with a glucose-responsive fluorescent monomer (GF-monomer) and supported by a PVA backing layer, without the use of enzymes or additional reagents [83]. The MNs were composed of two layers: a silk fibroin needle layer containing the GF-monomer, and a PVA backing layer for structural support. The GF-monomer consisted of a diboronic-acid moiety, which served as the glucose recognition site, and an anthracene moiety responsible for fluorescence emission. Upon application to the skin and excitation at 405 nm, the MNs sensor emitted fluorescence at 490 nm. In vitro studies showed that the MNs exhibited a greater fluorescence response in the hypoglycemic range than in the hyperglycemic range (50–450 mg/dL). The sensor was further evaluated across glucose concentrations ranging from 40 mg/dL (hypoglycemia) to 280 mg/dL (hyperglycemia), demonstrating high sensitivity and suitability for long-term in vivo glucose monitoring. The MNs sensor system supports convenient at-home glucose monitoring through a wireless portable device and a smartphone app with image-processing capabilities, and it maintains reliable performance even when applied at different skin angles (Figure 9B).

Previously, Zhang et al. incorporated fluorescent nanodiamonds into a fluorophenylboronic acid (FPBA)-functionalized hydrogel capable of generating fluorescence signals in response to changes in environmental glucose levels [80]. Notably, the team fabricated a fluorescent nanodiamond boronic hydrogel system in porous MNs using porogen leaching technology. The porous stature enhanced ISF extraction via capillary action, thereby reducing response lag time and improving in vivo glucose monitoring. The device exhibited good long-term photostability and signal stability in both small and large animal models.

8.5. SERS MNs Sensors for Glucose Detection

A SERS sensor allows for highly sensitive and quantitative detection of biomolecules by amplifying the Raman signals of molecules adsorbed onto plasmonic materials, such as gold (Au) or silver (Ag) nanoparticles or metal surfaces [130]. SERS enhancement originates from two principal mechanisms: the electromagnetic (EM) mechanism and the chemical mechanism (CM). The EM mechanism is driven by localized surface plasmon resonances (LSPRs), which amplify the local electric field around metallic nanostructures, while the CM mechanism involves direct charge transfer between the adsorbed molecule and the metal surface. Upon excitation, LSPRs induce collective oscillations of conduction electrons in metal nanostructures, generating highly intensified electromagnetic fields known as “hot spots” [131] The size, shape, and crystal structure of metal nanoparticles strongly influence the LSPR properties and, consequently, the magnitude of signal enhancement [132].

In composite SERS substrates, functional materials commonly include noble metals such as gold and silver, along with copper, metal oxides, metal-organic frameworks (MOFs), conjugated polymers, two-dimensional materials, and their corresponding composite structures [21]. Among these, Au and Ag are the most commonly used materials for bioanalytical SERS applications, including SERS-based MNs sensors. Silver nanomaterials exhibit excellent SERS enhancement performance and are widely used for ultra-sensitive in vitro analysis, although their potential toxicity remains a concern [133,134]. In addition to silver, Au nanomaterials demonstrate excellent biocompatibility and chemical stability, making them more suitable for in vivo and intracellular applications. There are various shapes of plasmonic materials and multicomponent nanoparticles, such as Au nanorods (Au NRs) [135], Au nanoparticles (Au NPs) [136], Au nano-islands (Au ILs) [86], and Ag nanoparticles (Ag NPs) [137], that have been incorporated into MNs-based SERS systems. The combination of MNs and SERS has been developed for the in situ monitoring of various biomarkers, such as glucose, offering a painless, effective, and minimally invasive strategy. The detected signals are further enhanced via the incorporation of metal nanoparticles.

Yuen and Liu were the first to fabricate an Ag-coated MNs-based SERS probe for the detection of Rhodamine 6G and glucose molecules embedded at depths exceeding 700 µm within a two-layer tissue phantom designed to mimic the optical absorption and scattering properties of human skin [126]. They demonstrated that glucose concentrations ranging from 5 to 150 mM within tissue phantoms could be quantified with a root mean square error (RMSE) of 3.3 mM. This work highlighted the potential of MNs for SERS-based measurements, enabling deeper tissue penetration and enhanced signal intensity. Nonetheless, the use of stainless-steel MNs induced a significant immune response in the skin.

In 2015, this research team further advanced a hollow agarose MNs coated with Ag for SERS detection of crystal violet and glucose molecules embedded at depths greater than 700 μm beneath the surface of a skin phantom [78]. The Ag-coated agarose MNs were able to measure glucose concentrations ranging from 5 to 150 mM within the skin phantoms, achieving a root-mean-square error of 5.1 mM in under 10 s. Moreover, Liu et al. reported a SERS sensor for in situ intradermal glucose detection based on a low-cost poly(methyl methacrylate) microneedles (PMMA MNs) array [75]. After functionalizing the silver-coated array surface with 1-decanethiol (1-DT), the sensor was calibrated over a glucose concentration range of 0–20 mM in skin phantoms. The functionalized PMMA MNs array was able to directly measure glucose levels in ISF of diabetic mice within a few minutes. Additionally, the MNs array demonstrated good recovery within 10 min after application, with no noticeable adverse reactions (Figure 10A).

Figure 10. MNs arrays employing SERS for transdermal glucose monitoring. (A) SERS based biosensor with a MNs array for minimally invasive in vivo glucose measurements [75]. (B) A porous SERS MNs for multiplex detection of glucose, uric acid, and cholesterol in interstitial fluid [65]. Adapted from Ref. [75] under the Creative Commons Attribution (CC BY 4.0) license.

In 2025, SERS probes and specific enzymes were integrated into porous MNs to enable the multiplexed detection of glucose, uric acid, and cholesterol [65]. Paraffin wax was used as a hydrophobic barrier to divide each porous MNs into three independent compartments. In these compartments, gold nanoshells (serving as SERS substrates), 4-mercaptophenylboronic acid (4-MPBA), and a specific enzyme were incorporated into each compartment. In this system, enzymatic reactions generate H₂O₂, which subsequently induces the conversion of substrate-immobilized 4-MPBA to 4-mercaptophenol (4-MPhOH), resulting in changes in the SERS signal. For glucose detection, the MNs platform demonstrated an in vitro detection range of 0–15 mM. The performance of the porous SERS MNs was further validated in rat models, confirming their capability for multiplexed in vivo detection (Figure 10B).

8.6. Photonic and Plasmonic Glucose Sensing

Photonic crystal MNs sensors rely on periodic nanostructures that selectively reflect specific wavelengths of visible light. When incorporated into glucose-responsive hydrogels, glucose binding can alter hydrogel swelling and consequently change the periodic spacing of the photonic structure. This shifts the reflected wavelength and produces a visible structural color change. Unlike conventional dye-based colorimetric systems, the resulting optical response originates from structural changes rather than consumption of a chromogenic reagent, potentially improving reversibility and enabling repeated measurements [23,138].

Zhang et al. developed a wearable MNs platform containing a PBA-functionalized photonic crystal hydrogel. Glucose binding altered hydrogel volume and shifted the reflected wavelength, allowing glucose concentration to be visualized by naked-eye color changes and quantitatively analyzed using smartphone images. The system demonstrated enzyme-free glucose sensing following ISF transport through the MNs platform and was evaluated using porcine skin as a human-skin model (Figure 11A) [138]. More recently, core–shell photonic crystal MNs have incorporated mechanically robust outer shells with PBA-responsive photonic hydrogel cores to improve skin insertion while preserving glucose-dependent structural-color sensing (Figure 11B) [139].

Figure 11. Representative photonic crystal MNs-based optical glucose biosensors. (A) Core–shell photonic crystal hydrogel MNs using phenylboronic acid–glucose recognition to induce hydrogel swelling, structural-color changes, and a glucose-dependent red shift in the reflection spectrum [139]. (B) Wearable transdermal MNs platform integrating a phenylboronic acid-functionalized photonic crystal hydrogel with smartphone-assisted color analysis for quantitative glucose detection [138]. Adapted from Refs. [138,139] under the Creative Commons Attribution (CC BY 4.0) license.

Plasmonic MNs biosensors instead exploit localized electromagnetic enhancement generated by metallic nanostructures such as gold nanorods or silver nanoparticles. In glucose sensing, the most established plasmonic MNs application is currently SERS, in which localized surface plasmon resonance enhances Raman scattering. Plasmonic paper integrated with MNs has previously demonstrated on-patch SERS analysis of molecules collected from ISF, establishing a general platform that subsequently supported glucose-specific SERS MNs development [140].

8.7. Smartphone-Assisted Optical Detection

Smartphones provide an attractive interface for optical MNs biosensors because their integrated cameras, processors, wireless connectivity, and user interfaces can convert visually observable signals into quantitative glucose measurements. In colorimetric systems, RGB values, hue, saturation, or luminosity can be extracted from images and correlated with glucose concentration. For photonic sensors, smartphone imaging can quantify structural-color changes or reflected wavelength shifts, whereas fluorescence-based systems can employ portable excitation modules together with camera or photodetector-based acquisition [138].

Sang et al. integrated their fluorescent biodegradable MNs sensor with a mobile application and portable optical accessories for tether-free CGM (Figure 12A) [83], while Zeng et al. used smartphone imaging to quantify colorimetric responses over 50–400 mg/dL [141]. Similarly, the photonic crystal MNs system reported by Zhang et al. used mobile-phone imaging to convert structural color into quantitative glucose information (Figure 12B) [138]. More recently, transparent chitosan-based MNs optical biosensors have also been coupled with Bluetooth-enabled readout electronics for real-time glucose measurements, highlighting increasing convergence between optical sensing, smartphones, and digital health technologies [142]. For reliable clinical use, however, smartphone-based optical quantification requires normalization for ambient illumination, camera model, viewing angle, skin pigmentation, sensor position, and optical background. Standardized illumination chambers, internal reference regions, ratiometric analysis, and computational image correction will therefore be important for reducing inter-device and inter-user variability.

Figure 12. Smartphone-assisted optical MNs platforms for glucose monitoring. (A) Biodegradable fluorescence-based MNs sensor integrated with a mobile application for image processing and continuous glucose monitoring [83]. (B) Photonic crystal-hydrogel MNs sensor coupled with smartphone-based color analysis for quantitative glucose determination [138]. Adapted from Refs. [83,138] under the Creative Commons Attribution (CC BY 4.0) license.

9. Advanced Materials for MNs-Based Optical Glucose Biosensors

Advanced functional materials play a central role in determining the sensitivity, selectivity, optical stability, response time, and biocompatibility of MNs-based optical glucose biosensors. Unlike conventional MNs that primarily provide access to ISF, optical MNs systems require materials capable of simultaneously recognizing glucose and converting this interaction into measurable fluorescence, colorimetric, Raman, photonic, or plasmonic signals. Recent developments therefore increasingly combine nanomaterials, glucose-responsive polymers, hydrogels, and multifunctional composites within a single MNs platform. Metallic nanoparticles, metal–organic frameworks (MOFs), fluorescent nanomaterials, photonic crystals, and stimuli-responsive hydrogels have been particularly important in advancing minimally invasive optical glucose sensing [14].

9.1. Metallic and Semiconductor Nanomaterials

Metallic and semiconductor nanomaterials provide unique optical properties derived from their nanoscale dimensions, including localized surface plasmon resonance, enhanced electromagnetic fields, catalytic activity, fluorescence, and wavelength-dependent light absorption or scattering. These characteristics can amplify weak optical responses and improve the detection of glucose within the small volumes of ISF accessible through MNs. Metallic nanostructures are particularly important in SERS and colorimetric sensing, whereas semiconductor nanomaterials can serve as fluorescent reporters, photocatalysts, or components of photonic structures [23].

Nanoparticles can also contribute structurally to optical MNs sensors. For example, ordered colloidal particles embedded in glucose-responsive hydrogels can create photonic band gaps whose reflected wavelength changes when the hydrogel swells. Liu et al. developed core–shell photonic crystal MNs consisting of a PBA-functionalized polyacrylamide hydrogel containing ordered SiO₂ nanoparticles and a mechanically robust hyaluronic acid shell. Glucose–PBA binding increased hydrogel swelling and lattice spacing, resulting in a concentration-dependent red shift of the structural color and reflection spectrum [139].

Although semiconductor materials such as ZnO, TiO₂, and semiconductor quantum dots are widely investigated in optical glucose biosensors, their direct integration into optical MNs glucose sensors remains considerably less developed than metallic, hydrogel, and photonic-crystal systems. Their high refractive index, fluorescence, catalytic properties, and ease of surface functionalization nevertheless make them attractive candidates for future MNs platforms, particularly when combined with PBA receptors or enzyme-mediated glucose recognition.

9.2. Gold and Silver Nanoparticles

Gold nanoparticles (AuNPs) and silver nanoparticles (AgNPs) are among the most important metallic nanomaterials for optical glucose sensing because of their strong plasmonic properties, large surface area, facile surface functionalization, and catalytic activity. Their localized surface plasmon resonance can enhance optical absorption and electromagnetic fields, while nanoscale Au and Ag surfaces generate intense SERS enhancement. These properties permit amplification of weak Raman signals and facilitate sensitive glucose detection at or near the MNs–ISF interface [14].

Ag has been particularly important for MNs-based SERS glucose sensing. Ag-coated polymeric MNs have been used as plasmonic substrates for direct detection of glucose in dermal ISF, demonstrating that MNs insertion can position SERS-active surfaces directly within the sensing compartment and overcome some limitations associated with transcutaneous Raman measurements. More recent porous SERS MNs architectures have integrated ISF transport, glucose recognition, and plasmonic signal enhancement within the same structure, further improving the potential for minimally invasive optical monitoring [14].

AuNPs can additionally function as nanozymes rather than solely as plasmonic materials. Li et al. developed swellable colorimetric MNs containing AuNPs with glucose oxidase-like activity. The AuNPs catalyzed glucose oxidation and H₂O₂ generation, while HRP subsequently mediated TMB oxidation to produce a visible color signal. The system enabled differentiation of normal and hyperglycemic states and allowed further quantification through digital image analysis [143]. This approach demonstrates how metallic nanoparticles can simultaneously contribute to glucose recognition, catalytic amplification, and optical signal generation.

9.3. Carbon Nanotubes and Graphene-Based Materials

Carbon nanotubes (CNTs), graphene, graphene oxide (GO), and related carbon nanomaterials possess high surface areas, tunable surface chemistry, excellent mechanical properties, and distinctive optical and electronic characteristics. In glucose biosensing, CNTs and graphene are predominantly used to enhance electrochemical rather than optical MNs sensing. For example, PBA-containing glucose-responsive polymers combined with CNTs have been coated onto MNs, where glucose-induced polymer swelling alters CNT electrical percolation and produces an electrical response [83].

Nevertheless, carbon nanomaterials also offer considerable potential for optical MNs biosensors. Graphene and GO are efficient fluorescence quenchers and can therefore function as acceptors in fluorescence-resonance energy transfer or fluorescence recovery systems. PBA-functionalized reduced graphene oxide has previously been demonstrated as an enzyme-free fluorescence glucose sensor, in which competitive glucose binding modifies fluorophore–graphene interactions and restores fluorescence [144]. Similarly, single-walled CNTs possess intrinsic near-infrared fluorescence that can be modulated by molecular recognition and has been investigated for optical glucose sensing [145].

However, direct integration of CNT- or graphene-based fluorescence glucose sensors into MNs platforms remains limited. Thus, this material class should presently be regarded as an emerging opportunity rather than an established optical MNs glucose-sensing strategy. Combining PBA-functionalized graphene, fluorescent CNTs, or graphene quantum dots with hydrogel-forming MNs may offer future routes toward highly sensitive, photostable, and reversible optical sensing.

9.4. Metal–Organic Frameworks

Metal–organic frameworks are porous crystalline materials composed of metal ions or clusters connected by organic ligands. Their exceptionally high surface area, tunable pore structure, catalytic activity, and ability to encapsulate enzymes or chromogenic reagents make MOFs particularly attractive for optical glucose biosensing. In MNs platforms, MOFs can function as enzyme carriers, catalytic nanozymes, or signal-amplification materials while maintaining analyte diffusion through their porous structures.

A direct example was reported by Wang et al., who developed Zr-MOF@Pt nanocomposites with peroxidase-like activity for visual glucose detection. GOx-mediated glucose oxidation generated H₂O₂, which was subsequently decomposed by Zr-MOF@Pt to oxidize TMB and generate a blue color. Incorporation of the Zr-MOF@Pt/TMB/GOx sensing system into a hydrogel MNs array produced a portable visual glucose assay, demonstrating a linear sensing range of 0.5–800 μM in the underlying colorimetric platform [91]. MOFs can also protect natural enzymes from environmental degradation. More recently, Li et al. incorporated both GOx and HRP within hydrophilic MAF-7 and embedded the dual-enzyme MOF system into hydrogel colorimetric MNs. The resulting platform combined ISF extraction with glucose detection across 1–25 mM, illustrating the potential of enzyme-loaded MOFs to improve catalytic integration within MNs-based optical systems [94]. However, MOF biocompatibility, metal-ion release, batch reproducibility, enzyme accessibility, and long-term stability require further investigation before prolonged in vivo use.

9.5. Quantum Dots and Fluorescent Nanomaterials

Quantum dots (QDs) and other fluorescent nanomaterials offer high brightness, narrow and tunable emission spectra, broad excitation ranges, and greater photostability than many conventional organic dyes. These properties are particularly attractive for MNs systems because low ISF volumes and limited optical path lengths require strong and stable fluorescence signals. Surface functionalization with glucose-recognition elements such as PBA or glucose-responsive enzymes can convert QDs into glucose-responsive optical probes. Graphene quantum dots and carbon dots are increasingly attractive because they may reduce concerns associated with heavy-metal-containing semiconductor QDs [146].

Among the most promising fluorescent nanomaterials for MNs glucose sensing are fluorescent nanodiamonds (FNDs). Their nitrogen-vacancy centers provide exceptional resistance to photobleaching and photochemical degradation. Zhang et al. incorporated FNDs into a PBA-functionalized hydrogel and coupled the sensing matrix with porous MNs. Glucose binding altered hydrogel swelling and consequently modulated FND fluorescence transmission, enabling reversible and prolonged glucose monitoring. The photostability of FNDs is particularly advantageous for continuous monitoring, where conventional fluorophores may progressively lose signal intensity [80].

Fluorescent glucose sensing can also be achieved without inorganic QDs. Hu et al. subsequently developed a wearable MNs patch incorporating reversible FRET-based hydrogel sensors, where glucose-specific PBA groups and polymerized FRET pairs generated concentration-dependent fluorescence changes. The FRET hydrogel achieved an LOD of 0.193 mM and a response time of 7.7 min in human serum [90].

10. Artificial Intelligence and Data-Driven MNs-Based Optical Glucose Biosensors

Artificial intelligence (AI) and data-driven analysis are emerging as promising approaches for improving the analytical performance of MNs-based optical glucose biosensors. These platforms generate diverse optical outputs, including fluorescence intensity, RGB color values, reflection-wavelength shifts, and Raman spectra, which can be affected by illumination, optical alignment, sensor-to-sensor variability, skin characteristics, temperature, and ISF volume. Digital preprocessing, such as background subtraction, normalization, region-of-interest selection, and feature extraction, is therefore important before glucose quantification. Current wearable optical sensing literature increasingly identifies AI and machine learning (ML) as useful tools for multidimensional signal analysis, calibration, and prediction, providing a foundation for intelligent MNs-based glucose monitoring [138,147].

AI and ML can contribute to MNs-based optical glucose monitoring at several distinct stages of the sensing pipeline. First, in signal processing, AI can support background subtraction, denoising, image segmentation, region-of-interest selection, spectral preprocessing, and automated extraction of optical features from fluorescence, colorimetric, photonic, or SERS signals. Second, multivariable calibration models can relate these features to glucose concentration while simultaneously accounting for confounding variables such as pH, temperature, illumination, skin characteristics, ISF volume, and sensor-to-sensor variability. Third, data-driven models can be used for drift correction by learning gradual changes associated with photobleaching, biofouling, enzyme degradation, substrate aging, or changes in optical alignment. Fourth, time-series models can analyze sequential ISF glucose measurements to estimate rates of change and predict future glucose concentrations or impending hypo- and hyperglycemic events. Finally, AI may assist physiological-lag compensation by modeling the temporal relationship between blood and ISF glucose together with device-specific response delays, thereby estimating a more clinically relevant glucose trajectory. These functions should be considered complementary components of the analytical pipeline rather than a single generic application of AI [148,149].

Smartphone-assisted optical analysis represents an important intermediate step between conventional optical sensing and fully AI-enabled glucose monitoring. Zhang et al. developed a PBA-functionalized photonic crystal hydrogel MNs platform in which smartphone-derived RGB values were converted into CIE chromaticity coordinates and subsequently related to glucose-dependent reflection wavelengths [138]. Sang et al. similarly integrated biodegradable fluorescent MNs with a mobile application for tether-free glucose monitoring. These systems primarily rely on conventional image processing rather than AI; however, recent smartphone glucose assays outside the MNs field demonstrate that convolutional neural networks (CNNs) can directly classify glucose-dependent colorimetric images while accounting for variations in cameras and illumination [83]. Kanchan et al., for example, reported a smartphone-based colorimetric glucose assay analyzed by a two-dimensional CNN, illustrating a computational strategy that could potentially be transferred to optical MNs platforms [150].

ML may further improve optical glucose quantification by modeling nonlinear relationships among multiple sensing and environmental variables. Potential model inputs include fluorescence intensity or ratios, RGB/HSV parameters, reflection wavelength, Raman spectral features, temperature, pH, sampling time, and sensor-specific characteristics. Algorithms such as partial least-squares regression (PLSR), support vector regression (SVR), random forest (RF), gradient boosting, artificial neural networks (ANNs), and CNNs could therefore support multivariable calibration and interference correction. This capability may be particularly important for PBA-functionalized photonic MNs because their optical response depends not only on glucose binding but also on hydrogel swelling, pH, ionic strength, and polymer characteristics. Liu et al., for example, showed that core–shell PBA-functionalized photonic crystal MNs exhibited both glucose- and pH-dependent reflection changes, highlighting the need for multivariable calibration rather than reliance on a single optical parameter [151,152].

AI may provide particularly substantial benefits for SERS-based MNs biosensors, because SERS measurements produce high-dimensional spectra with overlapping bands, baseline variations, background fluorescence, noise, and substrate-dependent signal variability. Conventional glucose analysis generally relies on selected characteristic peaks or peak-intensity ratios, whereas ML can simultaneously exploit information across an entire spectrum. PCA, PLSR, LDA, SVM, RF, ANN, and CNN approaches have been applied to Raman and SERS datasets for feature extraction, classification, and quantitative analysis. Deep-learning studies have further demonstrated that CNNs can distinguish structurally similar molecules from complex SERS signals, emphasizing the potential of this approach for biomolecular sensing [153,154]. However, these AI-SERS studies are not specific to glucose-sensing MNs; thus, their application to MNs-SERS glucose monitoring should currently be considered a promising future direction rather than an established technology.

Continuous optical MNs biosensors may additionally generate longitudinal datasets suitable for AI-based time-series analysis. Such models could use present and historical ISF glucose signals to identify glucose rates of change, compensate for sensor drift, distinguish transient artifacts from physiological variation, and forecast impending hypo- or hyperglycemia. Recurrent neural networks, long short-term memory networks, temporal convolutional networks, and related models have already been widely investigated using conventional CGM datasets. A systematic review and network meta-analysis of 46 studies demonstrated extensive application of neural networks, SVMs, RF, recurrent networks, and related models for glucose and hypoglycemia prediction, while AI-focused reviews of CGM identify sensor calibration, glucose prediction, and closed-loop control as major areas of application [148,149]. These established CGM frameworks could ultimately be adapted to compensate for physiological blood-to-ISF lag and optical sensor-response delays in continuous MNs systems.

Despite these opportunities, direct integration of AI with MNs-based optical glucose biosensors remains at an early stage, and most published optical MNs systems still employ conventional calibration equations, image processing, or spectral analysis rather than independently trained AI models. Major challenges include limited datasets, batch-to-batch differences in optical materials and MNs fabrication, variations in skin and ISF characteristics, ambient illumination, camera dependence, biofouling, sensor drift, and inter-individual physiological variability. Rigorous model development should therefore use subject- or sample-level separation between training and test datasets, independent external validation, standardized acquisition protocols, and prospective human evaluation. These considerations are especially important for SERS, where technical-replicate leakage and substrate batch effects can substantially inflate apparent ML performance [154]. Ultimately, integration of optical signals with physiological and temporal variables may enable AI-assisted self-calibration, drift and interference correction, physiological-lag compensation, and personalized glucose prediction, thereby transforming optical MN biosensors from passive detectors into intelligent glucose-monitoring systems.

11. Performance Evaluation of MNs-Based Optical Glucose Biosensors

11.1. Analytical-Performance Parameters

The performance of MNs-based glucose biosensors should be evaluated using standardized analytical parameters. Sensitivity represents the change in sensor response per unit change in glucose concentration and is generally obtained from the slope of the calibration curve. High sensitivity is particularly important for ISF analysis because only small sample volumes are available. Selectivity describes the ability to discriminate glucose from potentially interfering substances, including lactate, uric acid, ascorbic acid, acetaminophen, and other saccharides. Limit of detection (LOD) represents the lowest glucose concentration that can be reliably distinguished from background noise, whereas the Linear range defines the concentration interval over which sensor response remains proportional to glucose concentration. For glucose monitoring in ISF, the useful analytical range should ideally encompass both hypoglycemic and hyperglycemic concentrations [14,102].

Response time describes how rapidly the sensor reaches a defined fraction of its steady-state response after a change in glucose concentration and is particularly important for continuous glucose monitoring because sensor-related delay adds to the physiological blood-to-ISF lag. Stability reflects preservation of sensor response during storage and repeated or prolonged operation, whereas Reproducibility indicates the consistency of measurements among independently fabricated sensors or repeated analyses. For wearable MNs biosensors, these parameters should ideally be assessed not only in buffer but also in artificial ISF, ex vivo tissue, and ultimately in vivo conditions because biofouling, enzyme degradation, tissue response, temperature, and mechanical deformation can substantially alter performance [14,102].

11.2. Mechanical Characterization of Microneedle

Mechanical testing of MNs is essential to characterize their properties and to ensure their safety, efficacy, and overall quality [155]. Key evaluations include measuring the compression force of the MNs, the base-plate break strength and flexibility of the base plate, and the transverse-failure force of MNs arrays. Such assessments are also typically required during the regulatory review process. Therefore, these tests must be reliable, standardized, and reproducible. In addition, accurate evaluation of their physical dimensions and mechanical characteristics is critical, as these parameters play a vital role in optimizing device design and performance [156].

The dimensional characteristics of MNs must be carefully evaluated to ensure their safety and performance. Key parameters include needle length, height, inner and outer diameters (for hollow MNs), radius, and overall geometry. These dimensions can be accurately examined using scanning electron microscopy (SEM). Maintaining appropriate needle length and height is particularly critical, as excessive dimensions may penetrate too deeply and potentially damage the underlying layers of the skin [157]. Therefore, strict control of these parameters is essential during design and manufacturing. Additionally, the sterility of MNs arrays is a critical performance parameter, particularly for clinical applications. Ensuring the absence of bacteria, fungi, and other microorganisms is essential to prevent infection and to safeguard patient safety during use. This evaluation focuses on assessing the mechanical strength and structural integrity of the MNs. Important parameters include tip sharpness, tip geometry, and the ability of the MNs to withstand applied mechanical forces or pressure. MNs must possess sufficient mechanical strength to resist breakage during insertion into the skin. Ensuring adequate mechanical performance is essential to prevent device failure and to ensure safe and effective application [158].

11.3. Ex Vivo and In Vivo Performance

Ex vivo and in vivo evaluation is essential for determining whether MNs-based optical glucose biosensors can maintain reliable sensing performance under tissue-relevant and physiological conditions. Ex vivo studies commonly use porcine or rodent skin to assess MNs insertion, ISF transport, glucose recovery, and optical response. For example, Zhang et al. demonstrated glucose-dependent structural-color changes in a photonic crystal hydrogel MNs platform applied to glucose-equilibrated porcine skin, with results consistent with a commercial glucose meter, while Zeng et al. reported a strong linear response of porous colorimetric MNs in isolated rat skin over 25–400 mg/dL (R² = 0.9910) [138]. In vivo studies further assess dynamic glucose tracking, physiological lag, signal stability, skin recovery, and agreement with reference blood-glucose measurements. SERS-based MNs demonstrated direct glucose measurement in diabetic mice, with 93% of values located within zones A and B of the Clarke error grid, whereas biodegradable fluorescent MNs successfully tracked glucose changes after glucose and insulin administration in mice [83]. Fluorescent nanodiamond–boronic hydrogel MNs further enabled continuous glucose monitoring in mice and pigs, showing physiological lag times of approximately 3.8 ± 1.3 min during increasing glucose and 6.3 ± 3.3 min during decreasing glucose, together with prolonged in vivo sensing capability [80]. Overall, future validation should include standardized metrics such as correlation analysis, Bland–Altman analysis, Clarke or consensus error grids, lag-time assessment, mean absolute relative difference (MARD), prolonged wear, repeated-use safety, and ultimately prospective human studies to establish clinical reliability.

11.4. Biocompatibility and Safety

Biocompatibility and safety are critical considerations for MNs-based optical glucose biosensors because these devices directly contact viable skin and may remain in place for prolonged or repeated monitoring. Evaluation should include cytotoxicity, skin irritation, inflammatory response, tissue recovery, biodegradation, and possible accumulation or leaching of sensing components such as fluorescent probes, metallic nanoparticles, photonic materials, and glucose-responsive polymers. Similarly, fluorescent nanodiamond–boronic hydrogel MNs caused only mild and localized erythema after insertion, with significantly lower macrophage and T-cell infiltration than full-thickness wounds. In pigs, three-day continuous monitoring produced no significant skin inflammation, and the skin completely recovered within approximately seven days after device removal [80]. For photonic crystal MNs, Liu et al. reported cell viabilities of approximately 94–98% following exposure to the constituent materials, indicating favorable preliminary cytocompatibility. Nevertheless, long-term safety should also consider repeated insertion, chronic inflammation, foreign-body responses, phototoxicity, nanoparticle retention, degradation products, and systemic exposure to sensing materials [139]. Therefore, comprehensive in vitro cytotoxicity, histological evaluation, inflammatory-marker analysis, biodistribution, repeated-use studies, and long-term in vivo testing are required before optical MNs glucose biosensors can be translated into routine clinical monitoring.

12. Clinical Translation, Patents, and Commercialization

Wearable minimally invasive continuous glucose-monitoring (CGM) systems have substantially transformed diabetes management by providing real-time glucose trends and alerts that support therapeutic decision-making, particularly in individuals requiring insulin therapy. Several commercially available systems, including Dexcom G7^®, FreeStyle Libre^®, and Guardian Connect^®, measure glucose in interstitial fluid using subcutaneously inserted sensing elements and predominantly rely on electrochemical detection. For example, Dexcom G7 and Guardian systems use sensor filaments positioned within the subcutaneous ISF compartment. Optical glucose monitoring has also reached commercialization through Eversense 365, which employs a fully implantable fluorescence-based glucose sensor together with an external transmitter and mobile application; however, it is not a MNs-based device. In parallel, non-invasive Raman-based platforms such as MOGLU are under clinical development, illustrating growing interest in optical alternatives to electrochemical CGM, although such systems have not yet achieved the same commercial maturity as established CGMs. Importantly, MNs-based optical glucose biosensors have not yet been commercially launched, and most reported systems remain at the laboratory, preclinical, or early translational stage. Thus, currently marketed CGMs provide an important performance benchmark for future MNs optical technologies in terms of analytical accuracy, response time, operational lifetime, user comfort, wireless connectivity, and long-term reliability.

Nevertheless, increasing patent activity and early clinical investigations indicate substantial translational interest in MNs-based glucose monitoring. As summarized in Table 4, patented technologies encompass hydrogel MNs for electrochemical sensing and ISF extraction, aptamer-functionalized MNs biosensors, and optical platforms incorporating fluorescent nanodiamonds and glucose-responsive polymers. Human clinical investigations have also evaluated MNs or microprobe arrays for ISF glucose monitoring, indicating that the basic concept of minimally invasive MNs-mediated glucose sensing is clinically feasible. However, successful commercialization of optical MNs glucose biosensors will require several additional advances, including reproducible MNs fabrication and skin penetration, reliable ISF access, stable optical calibration, resistance to biofouling and environmental interference, long-term biocompatibility, sterilization compatibility, shelf stability, and scalable low-cost manufacturing. Optical systems must also demonstrate clinical accuracy against established CGMs or reference blood-glucose methods using standardized metrics such as MARD, Clarke, or consensus error-grid analysis, and lag-time assessment in sufficiently powered human studies. Therefore, although the growing patent portfolio, clinical investigations, and commercialization of non-MNs optical CGMs demonstrate a favorable translational pathway, a clear gap remains between current optical MNs prototypes and market-ready continuous glucose monitoring devices.

Table 4. Summary of commercialization status and patents related to MNs-based sensors for glucose detection in ISF.

Product/Technology	Development Status	Type of MNs	MNs Composition	Manufacturer/Institution, Country	Ref.
Dexcom G7® (electrochemical CGM)	Commercialized; ISF-based CGM	Non-MNs; subcutaneous filament sensor	—	Dexcom, USA	[159]
FreeStyle Libre® (electrochemical CGM)	Commercialized; ISF-based CGM	Non-MNs; subcutaneous flexible filament	—	Abbott Diabetes Care, USA	[160]
Guardian Connect® (electrochemical CGM)	Commercialized; ISF-based CGM	Non-MNs; subcutaneous sensor	—	Medtronic, USA	[161]
MOGLU (Raman spectroscopy)	Pre-commercial/clinical development; non-invasive optical CGM	Non-MNs	—	Apollon, USA	CES 2025
Eversense 365 (fluorescence-based CGM)	Commercialized; implantable optical CGM	Non-MN; implantable sensor	Fluorescence-based polymer sensor	Senseonics, USA	[162]
US20230113717A1	Patent application; electrochemical MN glucose/pH sensing	Hydrogel MNs	DA-HA/PEDOT-PSS with Pt and/or Ag NPs	University-affiliated inventors, USA/Canada	[163]
US20230321419A1	Patent application; ISF extraction and glucose analysis	Hydrogel MNs	Gelatin methacryloyl (GelMA)	University of California, USA	[164]
WO2025231565A1	Patent application; electrochemical aptamer biosensing	Hydrogel MNs	Methacrylated hyaluronic acid (MeHA) integrated with aptamer-functionalized electrodes	Canada	[114]
US20240374169A1	Patent application; optical glucose sensing	Hydrogel/porous MNs	Fluorescent nanodiamond + glucose-responsive boronic-acid hydrogel	USA	[165]
NCT02682056	Clinical trial; glucose measurement using MNs patches	Polymeric or metallic MNs patch	Biocompatible polymer or metal	Emory University, USA	ClinicalTrials.gov
NCT01908530	Completed clinical trial; microprobe-array CGM	Microprobe/MNs array (~1000 µm)	N.R.	Imperial College London, UK	ClinicalTrials.gov

Abbreviation: N.R.: Not Reported; Pt: platinum NPs: nanoparticles; PEDOT-PSS: poly(3,4-ethylenedioxythiophene) polystyrene sulfonate; DA-HA: dopamine-functionalized hyaluronic acid.

13. Current Challenges and Future Perspectives

The advancement of MNs sensing platforms has significantly transformed traditional invasive, blood-based diagnostics by enabling direct and continuous access to ISF, particularly for glucose monitoring. Throughout this review, we have demonstrated that integrating MNs with various optical sensing strategies provides minimally invasive and continuous glucose detection in ISF. Although several commercial continuous glucose monitoring systems are currently available, their high-cost limits widespread accessibility within the broader diabetic population. Meanwhile, researchers continue to develop innovative technologies for continuous glucose monitoring, aiming to enhance sensitivity, shorten assay time, reduce required sample volume, and improve overall analytical performance. However, several critical challenges must be addressed to achieve successful and widely adoptable continuous glucose detection in ISF. These challenges involve efficient functionalization with biorecognition elements (e.g., enzymes, antibodies, nucleic acids, and aptamers), efficient ISF extraction and sampling, high sensitivity and selectivity for low-abundance biomarkers, long-term biocompatibility and mechanical stability, robust sensor integration without structural compromise, reliable clinically relevant signal generation, and optimized optical performance with good skin compatibility [15].

At present, Raman methods, colorimetry, and fluorescence-sensing mechanisms have been integrated with MNs systems to enhance the development of MNs sensors and overcome the limitation of traditional blood-glucose monitoring. In future research, MNs-based optical sensing devices are expected to integrate wireless communication and connectivity capabilities, enabling continuous data transmission to smartphones or cloud-based platforms. This advancement will facilitate remote, real-time monitoring and feedback, thereby empowering users to make informed decisions regarding their health management. Furthermore, the integration of MNs-based optical sensors with artificial intelligence (AI) and machine-learning (ML) algorithms is anticipated to enable advanced data analysis and interpretation, supporting personalized detection and data-driven health recommendations. Machine learning can not only process complex spectral data to improve detection accuracy and resistance to interference, but also facilitate early disease warning and personalized health management through long-term continuous monitoring, pattern recognition, and trend prediction.

For clinical translation, future MNs-based optical glucose biosensors should be evaluated against quantitative performance benchmarks rather than analytical feasibility alone. Clinical accuracy should ideally approach that of established CGM technologies, with a MARD of approximately ≤10%, a high proportion of measurements within clinically acceptable Clarke or consensus error-grid zones, and a total blood-to-sensor lag preferably below 10 min. For wearable continuous monitoring, operational stability should extend from several days toward at least 7–14 days while maintaining calibration accuracy, optical signal stability, mechanical integrity, and biocompatibility [166]. Manufacturing translation will additionally require reproducible MNs geometry, sensing performance, and surface functionalization across production batches, together with scalable fabrication, sterilization compatibility, and acceptable shelf stability. Wireless systems should provide reliable real-time or near-real-time transmission to smartphones or external receivers throughout the intended wear period with minimal data loss and appropriate data-security safeguards. Finally, regulatory translation will require standardized analytical and clinical validation, biocompatibility testing, human-factor assessment, manufacturing quality control, and prospective human studies demonstrating safety and performance relative to established glucose-monitoring technologies. These targets should be regarded as practical translational benchmarks rather than universal regulatory thresholds [167,168].

In addition, further clinical investigations of MNs-based optical sensing systems for glucose detection are required to establish detailed correlations and validate their clinical reliability. Ensuring the biosafety of MNs sensors remains essential for long-term application. Moreover, the development of scalable and cost-effective manufacturing technologies is urgently needed to accelerate commercialization and broaden the practical deployment of MNs sensor arrays. Comprehensive evaluation of material and fabrication costs, as well as feasibility for industrial-scale production, is also necessary. From a SWOT (strengths, weaknesses, opportunities, and threats) perspective, MNs-based optical glucose biosensors possess several strengths, including minimally invasive access to ISF, the potential integration of sampling and sensing within a single wearable platform, compatibility with diverse optical modalities, and suitability for smartphone-assisted and data-driven analysis. Nevertheless, important weaknesses remain, particularly the limited and variable volume of accessible ISF, physiological blood-to-ISF lag, sensor-response delay, biofouling, signal drift, photobleaching, environmental interference, and insufficient long-term operational stability. Emerging opportunities include the integration of artificial intelligence and machine learning for multivariable calibration, interference and drift correction, glucose prediction, and physiological-lag compensation, together with wireless communication, cloud-based monitoring, advanced manufacturing technologies, and multiplex detection of additional ISF biomarkers. Conversely, major threats to clinical translation include competition from well-established commercial CGM technologies, stringent regulatory and clinical-performance requirements, batch-to-batch variability in MN fabrication and sensing materials, high manufacturing and validation costs, and unresolved issues related to sterilization, shelf stability, repeated-use safety, long-term biocompatibility, and user adherence. Collectively, this analysis indicates that future development should focus not only on improving analytical sensitivity, but also on achieving reproducible manufacturing, prolonged sensing stability, clinically acceptable accuracy, and robust validation under real-world conditions.