Sensor Technologies in Medicine–Food Homology: A Comprehensive Review

Abstract

Medicine–food homology (MFH) substances, which possess both medicinal and edible properties, have garnered widespread attention in the global health context of the new era. The MFH industry has experienced explosive growth and has gradually become a key supporting aspect of TCM modernization. However, due to the pollution of the modern environment, the content of pollutants in MFH products has been increasing, raising concerns regarding quality, safety, and efficacy control. Traditional quality-analysis technologies struggle to meet the needs of rapid on-site detection because of their dependence on large instruments and the complexity of operation. This dilemma has propelled advances in sensor technology. With its advantages of high sensitivity, real-time detection, and portability, sensor technology has become a key technical support for quality control and supervision in the field of MFH. In this review, we comprehensively categorize the mainstream sensor types used for analysis in the field of MFH, including intelligent sensors, optics, electrochemistry, biosensors, etc. This review outlines their research status, elaborates on their primary application directions and corresponding core technologies, discusses current challenges (including stability, interference, and cost), and presents future perspectives. Overall, sensor-based technologies offer a promising and scalable solution for the quality control of MFH products, addressing critical challenges such as stability, interference, and cost. With ongoing advances in intelligent sensing, optics, electrochemistry, and biosensing platforms, these methods are poised to play an increasingly vital role in ensuring the safety, efficacy, and quality consistency of MFH products amid growing environmental pressures.

1. Introduction

With the continuous advancement of the “Healthy China 2030” strategy and the deepening of the concept of “preventive treatment of diseases”, industries with the concept of big health as their core are ushering in unprecedented development opportunities [1]. In this context, as the crystallization of excellent traditional Chinese culture and modern scientific wisdom, the concept of “MFH” is becoming the core focus of today’s healthy consumer market, and its products are also favored because of their unique nutritional and medicinal values [2].

However, the surge in market demand has also brought a series of problems such as difficulty in exercising quality control, high safety risks, and a lack of effective means of authenticity identification [3]. Although traditional instrumental analysis methods are highly accurate, they face limitations such as high cost, long analysis cycles, and the need for laboratory environment. These methods make it difficult to meet the urgent needs of the industrial chain for the rapid screening of raw materials, online monitoring of the production process, and on-site market supervision [4].

Sensor technology meets the demands of the full-chain detection of MFH products with its unique advantages, including high sensitivity (predominantly for electrochemical and biosensors), excellent selectivity (prominent in optical and biosensors), rapid analysis speed, simple operation, and facile miniaturization and integration. At present, this technology has shown great application potential in many aspects, such as quantitative analysis of functional components [5], real-time monitoring of safety risk factors, and product authenticity identification and origin traceability [6]. Furthermore, its cross-integration with cutting-edge fields such as nanotechnology and artificial intelligence highlights the versatility and adaptability of smart sensor arrays in meeting contemporary challenges [7].

Based on this, this paper aims to systematically classify and review the latest research progress and application prospects of sensor technology in the field of MFH, discuss the application potential of sensor technology in the detection of functional components of MFH, safety monitoring and quality identification, and look forward to its future development directions, so as to promote the wider application of this technology in this field.

2. Overview of Sensors: Definitions, Principles, and Their Central Role in the MFH Field

A sensor is a detection device or system capable of perceiving a measured physical, chemical, or biological quantity and converting it into a standardized signal suitable for transmission, processing, and measurement [8]. Its operating principle involves an internal sensitive element measuring the target variable, which is then converted by a transducer into an electrical or other readily processable output signal according to a defined functional relationship [9]. The “sensing-conversion-output” process enables modern technology to meet comprehensive needs for data transmission, processing, storage, display, recording, and precise control. Sensor technology forms the cornerstone and source for data acquisition, status monitoring, and intelligent decision-making in modern measurement and control systems, the Internet of Things (IoT), and various intelligent devices.

3. Sensor Classification System: Systematic Categorization Based on Detection Principles and Operational Mechanisms

In the modernization research of MFH in traditional Chinese medicine, sensor technology is the core means to achieve precise quality control and intelligent process monitoring [10]. According to their detection objects and core working principles, sensors can be systematically divided into three categories: physical sensors, chemical sensors and biosensors (Figure 1). This classification system is the main framework discussed below.

3.1. Physical Sensors: Non-Destructive Monitoring of Environmental Parameters

Physical sensors are mainly used to detect physical environmental parameters, such as force, heat, light, etc., and their work does not depend on chemical reactions or biometric mechanisms. In the field of MFH, such sensors (e.g., mechanical sensors and temperature sensors) are widely integrated into intelligent manufacturing processes to enable real-time online tracking and control of extraction, drying, sterilization and other processes to ensure the uniformity of production conditions and stability. For instance, Catania et al. [11] designed an intelligent system in which temperature and humidity sensors, along with load cells connected to a wireless transmission module, continuously monitor moisture loss and drying rates during the drying process, thereby helping to ensure consistent product quality throughout the process. Furthermore, process analytical technology (PAT) frameworks incorporating physical sensors alongside spectroscopic tools have been established to enhance real-time process control in pharmaceutical and food manufacturing, improving both operational efficiency and final product consistency [12].

3.2. Chemical Sensors: Efficient Recognition of Chemical Components

Chemical sensors directly target the detection of chemical components. The sensitive element of chemical sensors interacts with target analytes through specific physicochemical processes, including surface adsorption, coordination complexation, redox reactions, and molecular recognition. For electrochemical sensors, the electrochemical reaction (e.g., redox reaction or ion exchange) of analytes on the surface of the sensitive electrode generates a measurable current or potential signal; for optical chemical sensors, the specific interaction (e.g., complexation, energy transfer, or conformational change) between the sensitive element and the analyte causes changes in optical properties (e.g., fluorescence intensity, absorbance), which are then converted into quantifiable optical signals [13]. Based on different signal transduction principles, they mainly include electrochemical sensors (e.g., potentiometric, amperometric) and optical chemical sensors (e.g., fluorescence, colorimetric/UV sensors). Both play significant roles in the quantitative analysis of active ingredients and the screening of harmful substances in Chinese herbal medicines. For instance, Oancea et al. [14] developed an amperometric microsensor based on graphite/carbon nanoparticles for the detection of quercetin in ginkgo essential oil, achieving a detection limit of 1.22 × 10⁻⁷ mol/L and a recovery rate of 97.4%.

3.3. Biosensors: High-Specificity Detection Based on Biological Recognition

Biosensors possess the highest specificity among the three categories. Their core lies in the use of biological recognition elements (e.g., enzymes, antibodies, nucleic acid aptamers) to achieve highly selective target recognition through specific biomolecular interactions (Figure 2). Antibodies bind to target antigens via antigen–antibody-specific recognition based on the complementarity of spatial structures; enzymes catalyze the specific reaction of corresponding substrates to produce detectable products; nucleic acid aptamers form stable three-dimensional structures to bind to target molecules through base complementary pairing and hydrogen bonding [15]. Notably, biosensors are defined by their recognition mechanism, while signal transduction relies on physical or chemical principles. Therefore, they are often further described based on the transducer type. For example, sensors that rely on antibody recognition and output a fluorescent signal can be referred to as immunosensors or fluorescent biosensors; similarly, if detection is based on the principle of surface plasmon resonance (SPR), the result is an SPR biosensor [16,17]. Biosensors come in a variety of sizes and forms and are capable of detecting low concentrations of pathogens, toxic chemicals, and pH levels [18].

At present, biosensor technology has been widely used in the fields of medicine, agriculture, pharmacy, and traditional Chinese medicine to determine a variety of analytes [19]. It has the advantages of strong specificity, mild detection conditions, and outstanding anti-interference ability. Common biosensors in the MFH field include enzyme sensors, nanosensors, immunosensors, nucleic acid sensors, and electrochemical biosensors. They enable rapid detection of active components, bacterial toxins, and pesticide residues. For example, Liu et al. [20] successfully synthesized a novel dual-recognition fluorescent biosensor (MPH) based on triphenylamine, which can be used for the rapid and reversible detection of Cu²⁺ and glyphosate pesticide residues both in vitro and in vivo. Dabhade et al. [21] developed laboratory-made micro-carbon electrodes and modified silver nanoparticles to construct a highly cost-effective aptamer sensor that could detect 34 CFU/mL of Escherichia coli within 15 min, providing a practical solution for large-scale screening at the grassroots level or in production areas. However, these methods still face challenges such as insufficient sensitivity, poor stability, and stringent environmental requirements. In particular, they are prone to interference from nonspecific adsorption in complex matrices, leading to an increased false-positive rate. Furthermore, certain biomolecular recognition elements, such as antibodies and enzymes, are susceptible to inactivation, which affects the long-term performance of the sensors. Therefore, there is a need to develop novel biomolecular recognition elements that combine high stability, strong resistance to interference, and ease of mass production [22].

The deep integration of multiple principles has significantly expanded the detection capabilities and application prospects of sensors in complex systems, such as the analysis of traditional Chinese medicine components. Based on this classification system, the following sections will provide a detailed discussion of the key sensors used in the field of MFH, along with their underlying principles.

Figure 2. Principle of biosensors. Created with BioGDP.com [23].

Figure 2. Principle of biosensors. Created with BioGDP.com [23].

4. Artificial Intelligence Sensory Technology

Artificial intelligence sensory is a bionic technology born in the 1960s, based on human senses and integrated with artificial intelligence to analyze its characteristics, including electronic tongue, electronic nose, and electronic eye. Artificial intelligence sensory technology has been widely used in the MFH field, and has been favored by scientists in the field of traditional Chinese medicine processing [24]. For instance, Carrillo et al. [25] used an electronic nose, electronic tongue, and electronic eye in combination with machine learning models to classify Colombian herbal tea brands; the best model achieved 100% classification accuracy, and the results were validated via HS-SPME-GC-MS analysis. Kallel et al. [26] used the α-Astree II electronic tongue system to evaluate the bitterness of five herbal tinctures and, in combination with principal component analysis, verified the masking effect of β-cyclodextrin on bitterness, providing experimental evidence for improving the palatability of herbal formulations. (Figure 3).

4.1. Electronic Eye (E-Eye)

The electronic eye is an intelligent sensory device that uses the principle of bionics to identify and analyze visual information. Based on colorimetry, spectrophotometry, or computer vision, it simulates human visual perception and collects and analyzes the shape, color, luster and other appearance properties of traditional Chinese medicine decoction pieces by simulating human visual perception. Parameters are collected and analyzed to realize the quality evaluation of different traditional Chinese medicines [27].

Compared with artificial vision, electronic eye technology has no visual fatigue, fast analysis speed, high accuracy, the availability of a large amount of information, and can improve the degree of automation and intelligence in actual production [28]. With the continuous development of science and technology, electronic eye technology has been applied in many aspects in the field of MFH. Through combined use with high-performance liquid chromatography, infrared, ultraviolet spectroscopy, etc., it can effectively realize the quality detection and evaluation of samples. Han et al. [29] combined electronic nose, tongue, and eye technologies with high-performance liquid chromatography (HPLC) and PLS-DA, SVM, and BP-NN models to identify Bletilla striata and its confusable species. In binary classification, PLS-DA and SVM models incorporating latent variables achieved 100% accuracy. In multi-class classification, PLS-DA based on fused components also performed best, with only one misclassification occurring during the validation phase. Key contributing factors included specific sensors, color features, and chromatographic peaks. Although this multi-technology fusion improved accuracy, it also increased data complexity and computational costs, posing challenges for practical applications. Shen et al. [30] realized the rapid, non-destructive and accurate evaluation of the processing amount of Epimedium by using electronic eyes and near-infrared spectroscopy combined with machine learning, providing a scientific basis for controlling the degree to which Epimedium and other traditional Chinese medicines are processed. However, this study did not account for the potential impact of environmental conditions (i.e., fluctuations in humidity and temperature) on the sensors. Since no drift correction strategy was employed in the study, its long-term stability remains in question.

4.2. Electronic Nose (E-Nose)

The electronic nose is an analytical instrument that simulates the mammalian olfactory system. It employs a gas sensor array (e.g., metal oxide semiconductors) that interacts with volatile organic compounds (VOCs) from MFH samples, generating measurable signals such as changes in resistance or frequency. These raw signals are then processed using techniques such as baseline correction and normalization before being analyzed by pattern-recognition algorithms (e.g., machine learning, linear discriminant analysis) for qualitative or quantitative analysis of unique “odor fingerprints,” enabling the identification of different odors and real-time monitoring of odor changes [31].

The rich volatile components (e.g., terpenoids, essential oils, aromatic derivatives) in traditional Chinese medicine are the material basis of their odor characteristics and important indicators of medicinal material efficacy [32], and their types and contents are closely related to the quality, origin, and processing of medicinal materials, which can be used as specific markers for electronic nose detection. Traditional methods involve judging quality according to the sense of smell. Compared with an artificial sense of smell, electronic nose technology can be combined with a variety of analytical instruments. Digital expression of odors can be realized [33]. In the context of smart manufacturing, electronic nose systems inspired by human olfactory mechanisms have attracted widespread attention in various fields such as food quality assessment, disease diagnosis, environmental factor monitoring, and safety. Electronic nose technology also mainly focuses on identification and processing in the field of MFH. By quantifying the odor properties, it is of great significance for the quality control and evaluation of medicinal materials [34]. Anisimov et al. [35] used an electronic nose based on organic field-effect transistors (OFETs) to detect spoilage gases when bacterial counts reached 4 × 10⁴ CFU/g—well below the safe consumption threshold—and to distinguish between different food types. Zhan et al. [36] used a self-assembled electronic nose system with 16 TGS sensors to distinguish 12 categories of traditional Chinese herbal medicines, including Astragalus, Glycyrrhiza, and Angelica sinensis. Eight features were extracted from each sensor, and multiple algorithms were compared. Among them, SVM achieved the highest identification accuracy of 98.94%, followed by LDA with 98.33%. PCA dimensionality reduction shortened classification time and enabled odor data visualization, while CP-1NN and CP-3NN provided reliable prediction confidence without sacrificing accuracy. This study offers a technical reference for herbal medicine flavor fingerprint library construction. In that study, they extracted 8 features from each sensor (such as maximum value, minimum value, slope, and integrated area), resulting in a total of 128 features from the 16 sensors. For classification tasks, high-dimensional features combined with a limited sample size may pose risks of feature redundancy and overfitting.

4.3. Electronic Tongue (E-Tongue)

Electronic tongue is an intelligent sensory analysis technology based on the principle of bionics, which responds to the overall chemical composition of liquid samples through a sensor array with low selectivity, non-specificity and interaction sensitivity, and combines signal processing and pattern-recognition algorithms to simulate human taste perception, thus realizing the qualitative or quantitative analysis of its taste characteristics [37]. Its working principle is mainly based on technologies such as potential analysis, impedance spectroscopy or voltammetry, and through pattern recognition of the response signal of the liquid to be measured, it can distinguish different tastes or evaluate taste intensity.

As a key component of TCM culture and dietary therapy, MFH products possess both food sensory attributes and medicinal functional attributes, requiring a balance of safety, functionality, and palatability [38]. Taste significantly influences consumer acceptance and market competitiveness. Current MFH food quality control research often lacks comprehensiveness and functionally targeted indicators. Consequently, e-tongue application for MFH quality control is gaining interest. Modern e-tongue technology can rapidly identify and digitally express basic tastes (sour, sweet, bitter, spicy, salty). For example, Lin et al. [39] used e-tongue data combined with a robust partial least squares (RPLS) regression method to accurately, safely, and efficiently assess the bitterness of traditional Chinese medicines, with the traditional human taste panel method (THTPM) as the gold standard for result validation. A high degree of consistency was observed between the e-tongue detection data and human sensory evaluation data: the RPLS model constructed based on e-tongue data achieved a cross-validation coefficient of determination (R_CV²) of 0.9394 and a robust root-mean-square error of cross-validation (RMSECV) of 0.3916 for bitterness values ranging from 0.63 to 4.78, and the model’s prediction performance showed negligible differences before and after outlier exclusion, which fully demonstrated the high correlation and reliability between the electronic tongue detection results and human sensory evaluation results. This study used the Traditional Human Taste Panel Method (THTPM) as a standard to validate the results obtained by the electronic tongue. However, human sensory evaluation has inherent limitations, such as high subjectivity, significant inter-individual variability, and poor reproducibility. Chen et al. [40] identified bitterness values of different ginsenosides by combining preparative HPLC with e-tongue and discovered bitterness-reduction methods, providing technical support for taste optimization of ginseng (an MFH product) and promoting MFH food standardization and functionalization. This study focused on how to identify and reduce bitterness, but did not examine whether reducing bitterness affects the medicinal efficacy of ginseng. For substances that are both food and medicine, the bitter components may in fact be the active ingredients.

5. High-Precision Sensors

Due to the complex composition of traditional Chinese medicine and environmental pollution, the content of microorganisms, toxins, pesticide residues and other components in traditional Chinese medicine has become a prominent problem. Simple sensory attribute analysis can no longer meet the detection needs, and high-precision sensing technology is needed to solve it. With the rapid development of modern analytical technology, high-precision sensors such as electrochemical sensors, biosensors, and fluorescent Sensors have been continuously developed and applied in MFH products, and are playing an increasingly important role.

5.1. Electrochemical Sensors

An electrochemical sensor is a measuring system that converts the response from an electrochemical reaction into a measurable electrical signal via a suitable transducer [41]. According to IUPAC, such sensors typically consist of a recognition element (receptor) that selectively interacts with the target analyte and a physicochemical transducer that converts this interaction into a signal. Based on the electrochemical measurement technique, electrochemical sensors are generally classified into potentiometric, voltammetric, amperometric, impedimetric, and conductometric types [42]. Potentiometric sensors measure the potential difference between an indicator electrode and a reference electrode; voltammetric and amperometric sensors monitor current as a function of applied potential; impedimetric sensors measure impedance changes, particularly charge transfer resistance; and conductometric sensors detect changes in solution conductivity [41,42]. The sensitivity and selectivity of these sensors are critically dependent on the composition and physicochemical properties of the electrode materials. Modifiers such as metal nanoparticles, metal oxides, carbon nanomaterials (e.g., carbon nanotubes, graphene), metal–organic frameworks (MOFs), polymers, and ionic liquids have been widely used to enhance electrocatalytic activity, increase electroactive surface area, and improve electron transfer rates, thereby enabling low detection limits, wide linear ranges, and good accuracy. Compared with conventional optical and chromatographic methods, electrochemical sensors offer advantages including simple and inexpensive instrumentation, minimal sample pretreatment, fast analysis, portability, and suitability for automation and real-time monitoring, making them particularly attractive for industrial quality control and on-site applications [43].

Electrochemical technology has the advantages of high sensitivity, high selectivity, fast detection speed and low cost, and has become a new and widely concerned quality control and quantitative analysis method in the MFH field [42]. They are widely used in TCM processing, authenticity identification, pesticide residue, mycotoxin, and heavy metal detection. For instance, Zhang et al. [44] developed an electrochemical sensor for baicalin determination using upconversion nanoparticles (UCNPs, typically NaYF₄: Yb, Er) and three-dimensional macroporous graphene (3DG). The UCNPs enrich target molecules via surface adsorption, while 3DG provides high conductivity and abundant active sites. Their synergistic effect enables sensitive detection with a linear range of 2.0 × 10⁻⁷ to 5.0 × 10⁻⁶ M and a low detection limit of 3.8 × 10⁻⁸ M. The sensor was successfully applied to determine baicalin in Scutellaria root samples. The synthesis of UCNPs in this experiment was complex, and the study only validates a single herbal matrix. For example, Argoubia et al. [45] developed a disposable electrochemical aptasensor based on ferrocene and aptamer-modified gold nanoparticles for the detection of ochratoxin A in food samples. The detection limit was as low as 11 pg/mL, and the sensor was successfully applied to coffee and wine samples without the need for complex pretreatment. This provides a valuable reference for the detection of ochratoxin A in food and medicinal products. Makani et al. [46] developed an acetylcholinesterase electrochemical biosensor based on Ti₃C₂T_X MXene quantum dots for the rapid detection of chlorpyrifos (a model OP), achieving a detection limit as low as 1 × 10⁻¹⁷ M. The sensor achieved low detection limits of 11.08 nM for Cd²⁺, 5.47 nM for Pb²⁺, and 6.42 nM for Hg²⁺, and was successfully applied to the analysis of representative herbal medicine samples. Mazaafrianto et al. [47] developed an electrochemical sensor for ochratoxin A (OTA) using a dithiol-modified aptamer, enabling rapid OTA detection in MFH traditional Chinese medicines. The sensor exhibited a wide linear range of 0.25 to 750 nM, a low detection limit of 113 pM, high selectivity against other mycotoxins, and was successfully applied to real samples, providing a basis for on-site detection and mycotoxin limit standard improvement. In this study, the stability of the dithiol-modified aptamer was not verified, and no data were provided regarding the number of times the sensor could be reused, its regenerative capacity, or its long-term storage stability.

5.2. Infrared Sensors

Based on the principle of molecular vibration spectroscopy, infrared sensors can detect the selective absorption of specific functional groups (such as O-H, C-H, N-H) at characteristic wavelengths, enabling non-destructive analysis of the chemical composition of samples. This technology has been widely used in the field of MFH, covering origin traceability, adulteration identification, component quantification and process control.

Near-infrared spectroscopy is widely used in the evaluation of MFH substances due to its speed and portability. Liu et al. [48] developed an ensemble learning algorithm enhanced by partial least squares discriminant analysis (Boosting-PLS-DA), combined with a portable near-infrared spectrometer, to identify the origin of MFH substances, and the accuracy rate of the external verification set reached 100%, effectively solving the problem of model overfitting. However, the signal strength of near-infrared spectroscopy is weak and the characteristic peak is wide. Its analysis accuracy is highly dependent on complex modeling and high training samples. The generalization ability of the model between different instruments or sample substrates is still a bottleneck that needs to be focused on in practical applications.

In terms of adulteration detection, handheld near-infrared spectroscopy combined with chemometrics has been used for rapid screening of honeysuckle [49], and its adulteration identification cross-validation accuracy reached 99.58%. This study reflects the significant advantages of near-infrared spectroscopy in field rapid detection, but due to its limited molecular specificity, it is often difficult to provide confirmatory results when distinguishing structurally similar components in complex matrices. At the same time, cloud-connected portable near-infrared technology can also be used for rapid screening of saffron [50], which utilizes a portable near-infrared spectrometer wirelessly connected to a smartphone via Bluetooth and combines it with cloud-based chemometric models to rapidly authenticate saffron on-site and predict adulterants with high accuracy (achieving 100% accuracy in authenticity verification using the PLS-DA model, with a minimum external prediction accuracy of 93% for adulterated samples), providing an efficient and feasible technical solution for on-site, non-destructive rapid testing of traditional Chinese medicinal materials. However, it suffers from issues such as cumbersome operation and limited applicability.

In contrast, mid-infrared spectroscopy has unique value in terms of molecular recognition specificity. Deconinck et al. [51] successfully realized the screening of specific medicinal plants in plant-based dietary supplements by using attenuated total reflection mid-infrared spectroscopy (ATR-FTIR) combined with SIMCA classification method, providing an effective technical means for the identification of adulteration in complex substrates. However, mid-infrared spectroscopy equipment has poor portability and has high requirements for equipment and operation. At present, it is still mainly used in laboratories, and it is difficult to meet the needs of high-throughput on-site screening.

5.3. Fluorescent Sensors

A fluorescent sensor is a bionic analysis technology based on molecular recognition and photophysical signal transduction. Specific binding between a recognition element and the target analyte induces photophysical property changes in a fluorophore, converting chemical recognition into a quantifiable fluorescence signal for highly sensitive and selective analysis [52].

At present, fluorescence sensing analysis technology has become a common method for quantitative analysis and detection, and is widely used in the quantitative analysis of toxic substances, pollutants, active substances, disease diagnostic markers and other components. The core of the accurate detection of fluorescent sensors lies in the process of “identification-signal conversion”, which can respond to specific substances through specific identification elements and quantitatively analyze the results. In the field of MFH, the common types of fluorescence sensors include metal–organic framework type, ion response type, nano-scale and molecular imprinted fluorescence sensors. It is mainly used to detect active ingredients, heavy metal content, mycotoxins, etc., in homologous substances of medicine and food. For example, Hu et al. [53] developed a fluorescent sensor array using MOF-scaffolded gold nanoclusters (AuNCs@MOFs) to accurately identify eight heavy metal ions (Ni²⁺, Cr³⁺, Co²⁺, Pb²⁺, Cd²⁺, Ag⁺, Cu²⁺, Zn²⁺) at 0.5–50 µM. Because heavy metal ions are non-biodegradable and may accumulate due to environmental pollution, posing a serious threat to human health through the food chain, their detection in traditional Chinese medicine is of paramount importance. By successfully distinguishing eight heavy metal ions in complex matrix samples such as Astragalus and Angelica sinensis, this study contributes to the quality control of traditional Chinese medicine and provides a reliable tool for assessing its safety. However, the preparation of these three types of AuNCs@ZIF-8 sensor elements requires a multi-step synthesis process, making the actual procedure relatively complex. Mu et al. [54] developed a novel fluorescent sensor based on glutathione functionalized graphene quantum dots (GQDs@GSH) for the detection of organophosphorus pesticide residues in radix Angelica Sinensis, exhibiting a good linear relationship with coumaphos concentrations in the range of 0.1–10.0 μmol·L⁻¹. Using the standard addition method, the recovery rates ranged from 101.44% to 117.90%, with a relative standard deviation (RSD) below 1.98%. However, the sensor developed by Mu et al. [54] was validated only for the detection of coumaphos and was not evaluated for its response to other common organophosphate pesticides. Given that traditional Chinese medicines may contain residues of various organophosphate pesticides, the practical applicability of this sensor would remain limited if it responds only to coumaphos. Conversely, if the sensor also responds to other organophosphate pesticides, its selectivity would be insufficient to distinguish which specific pesticide is present in the residues.

5.4. Surface Plasmon Resonance (SPR) Sensors

Surface plasmon resonance (SPR) sensors are an optical sensing technology based on the interaction between light waves and free electrons in metal films. The principle is that the change in the refractive index of the medium on the sensor surface will change the resonance conditions of the surface plasmon wave at the interface between the metal and the medium, thereby causing the shift in the transmission spectral vibration dip, and high sensitivity detection can be achieved by monitoring this shift [55].

In the field of MFH, optical fiber SPR sensors have attracted much attention because of their advantages such as miniaturization and remote detection, but their further development has long been limited by insufficient sensitivity and difficulty in achieving specific recognition. Wei et al. [56] proposed a wavy optical fiber surface plasmon resonance (SPR) sensor in their research to solve the problems of low sensitivity and difficult specificity recognition of traditional optical fiber SPR sensors in detecting active ingredients in traditional Chinese medicine, and realized the fast, highly sensitive, and specific detection of hyperoside, a component of traditional Chinese medicine. The sensor has the advantages of rapid production and high mechanical strength, which provides a new way for the rapid, high-sensitivity and high-specificity detection of active ingredients in traditional Chinese medicine. Klantsataya et al. [57] were the first to demonstrate SPR sensing in bare-core microstructured optical fibers. They used a chemical electroless plating technique to create a rough metal coating on fibers with a core diameter of only 10 μm, achieving a resolution twice that of conventional large-core fibers and a refractive index sensitivity of 1800 nm/RIU. This work provides a crucial foundation for the miniaturization and performance optimization of fiber-optic SPR sensors and offers valuable insights for the detection of food-grade ingredients. Addressing the need for rapid, accurate MFH product detection, Tang et al. [58] developed a ring-core fiber SPR sensor with high sensitivity and low detection limit, showing promise for miniaturized, portable TCM active-component detection devices. The successful development of these high-performance optical fiber SPR sensors is expected to promote the quality control of traditional Chinese medicine and the material basis research of drug efficacy into a new stage of rapid, accurate, and in situ detection, and has far-reaching significance for ensuring the safety and effectiveness of homologous products of medicine and food. However, in the three studies, special fiber structures such as wavy, spiral–conical, and ring-core fibers require highly sophisticated manufacturing processes, making them difficult for companies to implement in practice.

6. Applications in MFH Production and Research

The uniqueness of MFH products lies in the essential unity of medicinal efficacy and food safety [59]. Their overall quality depends on raw material quality, processing, active components, and safety indicators, imposing more complex and urgent demands for rapid, on-site quality control than ordinary herbs. Applying highly specific, intelligent sensor technology to four core stages—raw material identification, process monitoring, active component quantification, and safety risk screening—is a key trend in MFH quality control research. The following sections review sensor applications in these areas (Figure 4).

6.1. Raw Material Identification and Quality Control

The quality of medicinal and edible homologous materials will directly affect the efficacy and safety of the final product, so it is of great significance to identify and control the quality of its raw materials. So far, the development of quality assessment of Chinese medicinal materials (CMC) has gone through four stages: offline analysis, online analysis, online analysis, and on-site analysis [60]. Today, new sensor technologies can provide more sensitive, convenient, and effective means compared to traditional technologies [61], playing significant roles in MFH raw material identification and quality control, primarily using AI sensory technologies, fluorescent sensors, and nanosensors.

AI sensory technologies include e-eye, e-nose, and e-tongue [62]. An electronic nose (E-nose) can overcome the challenge of complex odors and difficult specific discrimination by identifying the odors of Chinese herbal medicines [63]. The electronic tongue can be digitally expressed by distinguishing different tastes. The electronic eye can establish a suitable data discrimination model by collecting color data. Examples include the work of Jin et al. [64], who realized Astragalus traceability detection by combining an electronic tongue and electronic eye with a new method involving a lightweight convolutional neural network transformer model. Yang et al. [65] used an electronic nose in combination with three machine learning methods to achieve an accuracy rate of over 90% in identifying the origin of Atractylodes macrocephala rhizomes (PLS-DA: 96.88%, BPNN: 96.88%, PSO-SVM: 100%). However, the validation method used to achieve the 100% accuracy is questionable and may indicate overfitting. Guo et al. [66] used electronic tongue technology to obtain sweetness values, used a colorimeter to determine yellowness values, and captured odor fingerprint information through the electronic nose (E-nose). The accuracy of predicting the content of alcohol-dissolved extract and polysaccharide extract, as well as the taste and color of Codonopsis pilosula can reach more than 85%. However, to further improve the predictive accuracy of this model, future research should utilize a larger sample size and incorporate more sophisticated feature extraction techniques, as well as optimize the machine learning algorithms.

Fluorescent sensors combine materials with different fluorescent properties, converting physical changes into signals for analysis [67]. Examples include the work of Long et al. [68], who used a sensor array of gold nanoclusters (AuNCs) and quantum dots (QDs) to identify the origins of Lilium bulbs (BH); this sensor array clearly identified the origin of BH with a prediction accuracy of 94.4%. Bian et al. [69] used a novel 2 × 3 six-channel fluorescent sensor array with machine learning and an indicator displacement assay to rapidly distinguish honeysuckle (Lonicerae japonicae flos) from Japanese honeysuckle (Lonicerae flos) in MFH. Although this method achieved an accuracy rate of only 91.50% in distinguishing among the four phenolic acids, it achieved 100% accuracy in distinguishing between Japanese honeysuckle flower (Lonicerae japonicae flos) and honeysuckle flower (Lonicerae flos); further research is needed to determine the reason for this.

A detailed comparison of sensor technologies for raw material identification and quantitative analysis of active components is provided in

Table 1. Quality and authenticity evaluation of traditional Chinese medicine.

Application Domain	Research Object/Target	Core Technology/Methodology	Limitations	LOD/Sensitivity	Ref.
Raw Material Identification & Quality Control	Atractylodes macrocephala (Baizhu)	Electronic nose integrated with three machine learning approaches	Physicochemical indicators are unstable and cannot distinguish origin; color parameters are greatly affected by processing methods; classification accuracy below 80%	100% accuracy	[65]
	Lilium (Baihe)	Gold nanocluster and quantum dot sensor array	Batch-to-batch stability not verified; generalization ability to new origins needs validation; long-term sensor stability not reported	Successful discrimination	[68]
	Lonicera japonica (Jinyinhua) and Lonicera macranthoides (Shanyinhua)	2 × 3 six-channel fluorescent sensor array	Accuracy for distinguishing single phenolic acid components is only 91.50%; generalization ability to more products from unknown manufacturers not verified; applicability of sample pretreatment method in different matrices needs confirmation	Rapid differentiation	[69]
Quantitative Detection of Active Constituents	Crataegus pinnatifida (Shanzha)	Smartphone-assisted dual-mode fluorescent sensor	Deacidification treatment leads to loss of functional components (e.g., flavonoids); high organic acid content requires addition of large amounts of sugar for flavoring, limiting use by elderly and diabetic patients; lack of products that preserve functional components while improving taste	Sequential detection of Fe3+ and vitamin C	[70]
	Baicalin	CS/ACK@CeO2-NPs composite electrochemical sensor	Low bioavailability and poor gastrointestinal absorption are major challenges; blood–brain barrier penetration is controversial; lack of clinical research, safety not fully established	4.81 × 10−9 mol/L	[71]
	Epimedium (Baohuoside I)	Graded-index multimode fiber SPR sensor	Sensitive to temperature changes, requiring constant-temperature environment; fabrication involves precise fiber processing (tapering, grooving, coating), resulting in relatively high cost; long-term stability of targeted protein immobilization on fiber surface needs verification	0.66 nm/(μg/mL); LOD: 0.15 μg/mL	[72]

. Key challenges include batch-to-batch stability, generalization across origins, and long-term sensor reliability.

6.2. Processing Monitoring

6.2.1. Preprocessing and Cleaning

Roots, stems, flowers, fruits, seeds, and leaves of MFH materials are prone to physical impurities (stones, soil, metal, plastic) during harvesting, affecting purity, sensory quality, and causing equipment wear. Some MFH substances also have toxicity or hardness, requiring preprocessing and cleaning [73]. The traditional method of relying on manual picking is inefficient, shows poor stability, and struggles to meet the needs of large-scale industrial production. Therefore, sensor-based intelligent online sorting equipment has become a standard configuration on modern production lines. It mainly uses non-destructive testing technologies such as machine vision and X-ray to achieve accurate and efficient removal of impurities.

In machine vision-based inspection systems, high-resolution cameras capture real-time images of materials moving along the production line. These images are transmitted to an embedded processor or computer, where algorithms compare color, shape, texture, and size against preset thresholds derived from the visual “fingerprints” of acceptable materials. Based on this analysis, the system rejects defective grains exhibiting discoloration, mold, or spots (e.g., in the sorting of goji berries or coix seeds) [74]. However, this method is inherently limited to surface inspection: it cannot detect impurities embedded within the material, nor can it identify impurities that resemble the target product in color and shape, and it is unable to distinguish between items based on density differences.

Secondly, the detection technology based on X-ray transmission provides an irreplaceable solution for identifying impurities with a density similar to that of raw materials. It is difficult for machine vision to effectively distinguish impurities hidden inside materials or similar in color and shape. X-ray detection technology forms gray-scale contrast images based on the difference in X-ray absorptivity of different substances. High-density impurities such as metal, glass, and stones will present obvious dark shadows, and even if they are wrapped inside the raw materials, they can be recognized by the system [75,76]. In the traditional Chinese medicine or food industries, when dealing with raw materials derived from both medicinal and food sources (which often have variable densities and irregular shapes), where it is difficult to distinguish foreign objects or tissue variations within the material, x-ray phase contrast imaging (XPCI) can be used to improve the imaging contrast of low-absorption substances, thereby aiding in the identification of traditional Chinese medicines. This technology is particularly useful for density-variable, irregular MFH materials like roots and fungi, complementing machine vision.

Such automated preprocessing enhances MFH product safety and appearance uniformity and provides a clean, uniform material basis for subsequent intensive processing (e.g., Processing (Paozhi), extraction).

6.2.2. Processing (Paozhi)

Processing is the core link in shaping the efficacy and flavor of MFH substances, and different processing methods can affect the chemical composition and proportion of MFH substances [77]. The success or failure of the traditional processing process is highly dependent on the mastery of “heat” (a traditional concept referring to the mastery of temperature and duration). In this process, pharmacists have long judged the endpoint by observing color changes with the naked eye, smelling and odor changes, and differences in feel and texture. It is full of subjectivity and uncertainty, which not only leads to fluctuations between batches of product quality, but also restricts its modernization and large-scale development [78]. Sensor technology is transforming control from experience-based to data-driven by objectively monitoring physical and chemical changes in real time. Intelligent sensory sensors (e-eye, e-nose, e-tongue) and process analytical technology (PAT) like NIR spectroscopy enable real-time, non-destructive monitoring of color, odor, taste, and key chemical dynamics.

In a study on ginger-processed Magnolia officinalis, a colorimeter, electronic nose, and electronic tongue were used to quantify the sensory differences between the samples before and after processing. Combined with chemometric models, these quantitative data were correlated with 26 distinct chemical components, enabling effective differentiation between unprocessed and ginger-processed Magnolia officinalis. Furthermore, subsequent validation through animal or cell experiments could confirm whether the toxicity of the processed samples is indeed reduced, thereby providing scientific evidence for the traditional belief that “ginger processing reduces toxicity” [79].

Process analysis technology (PAT) represented by near-infrared (NIR) spectroscopy and hyperspectral imaging (HSI) is committed to in situ and non-destructive online monitoring of key internal substances such as moisture and active ingredients during the processing process, so as to achieve real process feedback and precise control. Near-infrared spectroscopy and other technologies can monitor the real-time changes in moisture and active components in honey processing online, which provides the possibility of realizing closed-loop feedback control in the production process [80]. Fu et al. [81] used FT-NIRS to build rapid prediction models for nutrients (soluble sugars, proteins) in Lanxangia tsaoko, marking a shift from identification to quantitative tracking of specific components, providing a key tool for process optimization and standardization. Data fusion enhances reliability; e.g., combining NIR with laser-induced breakdown spectroscopy (LIBS) simultaneously acquires organic and inorganic information, comprehensively assessing complex physicochemical changes during processing.

6.2.3. Formulation and Packaging

The formulation and packaging of MFH products is the last key process to ensure stable efficacy, safe food, and an excellent experience. The deep integration of modern sensor technology can realize the control of the preparation and production process and the intelligent monitoring of the packaging link. Sensor technology improves quality in the final stage of the production of MFH products, and at the same time allows for consumer-friendly packaging, fully guaranteeing the medicinal reliability and food safety of MFH products.

In the preparation and production process, ensuring the uniform content of active ingredients and stable physical and chemical indicators is the basis for achieving product batch consistency. Process analysis technology realizes real-time and non-destructive monitoring of key production parameters by integrating online sensors such as near-infrared spectroscopy and Raman spectroscopy. “Raman plus X” dual-modal spectroscopy, combining Raman and NIR, shows unique advantages [82]. Raman spectra are highly sensitive to changes in molecular functional groups and crystal forms, which can effectively identify the authenticity of components and monitor the reaction process; near-infrared spectroscopy is good at rapid quantification of macroscopic components such as water and total sugars. Combining the two and using data fusion modeling can accurately capture and predict a wider range of chemical and physical changes in the preparation process, laying a technical foundation for the establishment of an online intelligent quality control system covering more comprehensive quality attributes.

The core of smart packaging is to be able to generate a perceptible response to external or internal ambient temperature, specific gases, microbial metabolites and other stimuli [83]. Siciliano et al. [84] reviewed optical gas sensors for smart food packaging and noted that colorimetric sensors are practical for detecting indicators of spoilage. These sensors convert chemical changes into visible color signals, allowing consumers to assess product freshness without specialized equipment. However, their application in MFH products is limited by the following factors: poor stability under fluctuations in light, temperature, and humidity, as well as their semi-quantitative nature and susceptibility to gas interference. In contrast, while electrochemical sensors offer higher sensitivity and quantitative accuracy, they require complex instrumentation and are therefore less suitable for consumer use. Consequently, colorimetric sensors are better suited for low-cost visual screening, whereas electrochemical methods are more appropriate for laboratory quality control.

6.3. Quantitative Detection of Active Components

The core value of MFH substances lies in medicinal and food functions. Medicinal value primarily resides in active components like flavonoids, saponins, polysaccharides, and alkaloids, proven to have various physiological activities (antioxidant, antibacterial, antiviral, anticancer, anti-inflammatory) [85]. Multiple steps (cultivation, transportation, processing, storage) can affect active component content, impacting final quality [86]. Sensor-based detection has become an emerging method for rapid TCM active-component detection. In the MFH field, sensors for quantitative detection mainly include fluorescent, electrochemical, and fiber-optic SPR sensors.

With the introduction of fluorescent nano-technology, fluorescent sensors are gradually applied in the MFH due to their advantages of high sensitivity and high selection. For example, Yeganeh-Salman et al. [87] developed an OFF–ON fluorescence sensor based on Schiff’s base and nanostructured molecularly imprinted polymers for the detection of rutin residues in fruit juices, with a linear range of 1–200 μg/L and a detection limit as low as 0.024 μg/L. The method has been validated in raspberry and green tea samples, but has not yet been tested in complex matrices such as traditional Chinese medicine formulations or biological samples; its ability to resist matrix interference in food–medicine products requires further evaluation. Huang et al. [70] developed a dual-mode fluorescent sensor with smartphone-assisted analysis, using concentration-dependent fluorescence changes for sequential Fe³⁺ and vitamin C detection in hawthorn, potentially advancing modernized TCM component detection.

Electrochemical analysis technology is increasingly used to determine active ingredients in traditional Chinese medicine due to its rapid response, accuracy and high sensitivity [88]. Examples include the work of Sun et al. [71], who developed an electrochemical sensor based on a CS/ACK@CeO₂-NPs composite for the detection of baicalin—a key quality indicator of the anticancer herb Scutellaria baicalensis—with a detection limit as low as 4.81 × 10⁻⁹ mol/L. The primary advantage of this sensor lies in its material design, which enhances electron transfer and increases the effective surface area, thereby offering higher sensitivity and a broader linear range compared to fluorescence-based methods. However, its limitations include a complex, multi-step preparation process, which poses challenges for scalability and reproducibility. Furthermore, the sensor has only been validated in simple herbal extracts and has not yet been tested in complex matrices. Therefore, while the sensor is well-suited for laboratory quality control, its practicality for point-of-care testing or field applications is limited. Liu et al. [89] used cinnamon residue to prepare mesoporous carbon for a rutin electrochemical sensor (LOD 11.7 nM, wide linear range 0.05–1 µM and 1–40 µM, excellent repeatability/stability). This is significant for rutin detection in flavonoid-containing MFH drugs (e.g., hawthorn, pagoda tree flower, buckwheat) and adds value to cinnamon residue utilization.

The fundamental/core principle of surface plasmon resonance (SPR) for analyzing biomolecular interactions involves immobilizing one target biomolecule on a sensor chip while allowing another molecule to pass over the surface. When the second molecule binds to the immobilized molecule, the surface refractive index changes, causing a shift in the SPR signal. By monitoring this change, researchers can determine the interaction strength and other properties such as kinetics and affinity constants [90,91]. For example, Minunni et al. [92] provided a systematic overview of the application of surface plasmon resonance (SPR) biosensors in the screening of bioactive compounds from plant extracts; using extracts from the prickly ash tree as an example, they demonstrated the experimental workflow for screening bioactive compounds that interact with DNA using SPR. Wei et al. [72] proposed and demonstrated a novel fiber-optic SPR sensor with high sensitivity and specificity by adopting graded-index multimode fiber as the sensing fiber. The sensor exhibits a detection sensitivity of 0.66 nm/(μg/mL) and a limit of detection (LOD) of 0.15 μg/mL, enabling the specific recognition of Baohuoside I—an active component in the traditional Chinese medicine Epimedium.

6.4. Quality and Safety Screening

Increased global pollution and pesticide use have raised concerns about heavy metal contamination, pesticide residues, and fungal contamination in MFH products during cultivation, harvesting, processing, transportation, and storage [93]. Although numerous studies have examined the active ingredients and pharmacology of MFH, safety assessments remain relatively scarce [94]. The food health industry faces challenges, particularly in quality standards research, while food attribute research remains in its infancy [95]. Reliable, sensitive detection technologies are urgently needed for safety evaluation and standard establishment. Sensor technology, with its multi-dimensional data analysis capability, enables detection of heavy metals, pesticide residues, fungal contamination, etc., based on different principles, becoming a powerful tool with broad prospects.

Heavy metals are widely present in nature and are not biodegradable. The use of heavy-metal-containing pesticides and increasing environmental pollution has led to the accumulation of heavy metal residues. Consequently, heavy metals such as cadmium (Cd²⁺), mercury (Hg²⁺), lead (Pb²⁺), copper (Cu²⁺), and arsenic (As³⁺) in traditional Chinese medicines have become a common phenomenon [96]. Heavy metals pose a serious threat to human health, and since most food–medicine dual-use products are ingested directly, their risk assessment techniques have drawn global attention [97]. Mercury (Hg²⁺) damages the brain, nervous system, liver, and kidneys even at low concentrations. For example, Eksin et al. [98] developed an environmentally friendly electrochemical sensor for the rapid detection of Hg²⁺ using a graphite pencil electrode modified with herbal silver nanoparticles; the detection limit was 8.43 μM, and the detection time was only 1 min. However, TCM samples contain macromolecules such as polysaccharides and tannins, which tend to adsorb onto the electrode surface, causing signal decay and poor repeatability. Therefore, industrial application requires anti-fouling electrode interfaces or standardized pretreatment protocols. Lead (Pb²⁺) is also highly toxic. Tu et al. [99] designed a simple, rapid, sensitive peptide-modified nanochannel sensor based on a Pb²⁺-specific peptide-modified porous anodic alumina membrane (PAAM) for accurate Pb²⁺ detection in TCM samples, enabling accurate Pb²⁺ detection in MFH substances. Nevertheless, particulates or macromolecules in complex TCM matrices can clog the nanochannels, while the porous anodic alumina membrane may suffer structural degradation or peptide detachment after repeated use, compromising long-term stability and regeneration capability.

Bacterial toxin contamination primarily occurs during the processing, transportation, and storage of raw materials, posing a hidden and extremely high-risk safety threat in substances with dual medicinal and food applications [100]. With widespread contamination, detecting bacterial toxins like aflatoxin B1 (AFB1) and ochratoxin A (OTA) in MFH substances is increasingly important. Aflatoxin B1 (AFB1) is a highly dangerous carcinogen easily contaminating food and TCMs, raising dietary health concerns [101]. Liu et al. [102] fabricated a novel ultra-high sensitivity electrochemical aptamer biosensor (EAB) by immobilizing gold nanoparticles (AuNPs) at the nanoconfined interface of N-doped carbon nanofibers/carbon fibers (N-CNFs/CFs) for immobilization and rapid, efficient AFB1 detection in complex matrices. However, the stability of the nano-confined interface under repeated use and the potential for matrix-induced fouling in complex traditional Chinese medicine decoction pieces remain to be addressed before industrial application. Liu et al. [103] used OTA-triggered antiparallel G-quadruplex interaction with (N-methyl-4-pyridyl) porphyrin (TMPyP) for rapid, sensitive OTA determination, showing high selectivity over similar mycotoxins, successfully applied to Astragali Radix samples. Jamal et al. [104] developed an electrocatalytic aptamer sensor based on an adapter/thiol-functionalized PEG-modified gold wire screen-printed electrode via hybrid self-assembly, achieving signal amplification through methylene blue-mediated ferrocyanide reduction. This sensor was used for the rapid detection of E. coli, with a detection range of 10–1000 CFU/mL and a detection time of only 30 min, and it was not interfered with by Bacillus subtilis.

Pesticide residues in MFH substances are a core safety concern from production to consumption, relating to both “medicine” safety/efficacy and “food” long-term health [105]. In the “great health” era, focus on healthcare and environmental monitoring has greatly influenced sensor development for rapid pesticide residue detection, particularly optical, electrochemical, and fluorescent sensors [106]. Chlorpyrifos, an insecticide/acaricide, poses serious human hazards. Sun et al. [107] constructed an electrochemical enzyme biosensor using acetylcholinesterase for highly sensitive, rapid chlorpyrifos detection in TCMs (LOD 7.90 × 10⁻⁵ ppm within linear range), showing stability, reproducibility, and sensitivity, enabling rapid on-site detection. Methyl parathion, an organophosphate insecticide, can leave residues affecting TCM quality with frequent use [108]. Sun et al. [109] fabricated a rapid, simple enzyme biosensor (AChE/AuNPs/PB/CS@N-Gr/GCE) for sensitive methyl parathion detection (LOD 9.47 × 10⁻⁵ µg/mL within linear range), providing a rapid, simple method for trace detection and portable on-site use (

Table 2. Safety risk factor screening of traditional Chinese medicine.

Application Domain	Research Object/Target	Core Technology/Methodology	Limitations	LOD/Sensitivity	Ref.
Safety Screening: Heavy Metal Detection	Heavy metal ions (eight species)	MOF-supported gold nanocluster fluorescent sensor array	Fluorescence quantum yield of gold nanoclusters is low; stability of MOF in complex matrices requires further validation; sensor fabrication reproducibility needs evaluation	0.5–50 μM range	[53]
	Pb2+	Pb2+-specific peptide-modified porous anodic alumina sensor	Although selectivity was verified for 11 interfering ions, unknown interferences may exist in more complex matrices; preparation involves multiple chemical modification steps, making the process rather complicated; reusable over 50 cycles but degree of performance decline not specified	0.1 ppb; >50 cycles	[99]
Safety Screening: Mycotoxin Detection	Aflatoxin B1 (AFB1)	N-doped carbon nanofiber/carbon fiber electrochemical aptasensor	Batch-to-batch reproducibility of electrospun nanofibers needs evaluation; long-term storage stability of the sensor not verified	Rapid and efficient detection	[102]
	Ochratoxin A (OTA)	OTA-triggered antiparallel G-quadruplex fluorescent sensor	Coexisting substances such as nickel ions, mercury ions, and pesticides in complex environmental samples may quench CdTe QDs fluorescence, affecting detection accuracy; applicability in more actual sample matrices needs verification	High selectivity	[103]
Safety Screening: Pesticide Residue Detection	Chlorpyrifos	Acetylcholinesterase electrochemical biosensor	Biological enzymes have poor stability and insufficient tolerance; enzyme inhibition techniques may produce false positives	7.90 × 10−5 ppm	[107]
	Glyphosate/Cu2+	Triphenylamine-based dual-recognition fluorescent biosensor	Continuous detection relies on “turn-off-on” sequence; the two analytes may interfere with each other when present simultaneously; detection performance in real TCM samples needs further verification; triphenylamine-based fluorescent probes may have photobleaching issues	Rapid and reversible detection	[20]
Safety Screening: Microbial Detection	Escherichia coli	Silver nanoparticle-modified microcarbon electrode aptasensor	Selectivity against other bacterial species needs further verification; long-term stability of silver nanoparticles; applicability in more complex matrices such as TCM extracts needs validation	34 CFU/mL; 15 min	[21]

).

However, as enzyme-based biosensors, their long-term storage stability, resistance to matrix interference from complex herbal samples, and selectivity against structurally similar organophosphorus pesticides remain critical challenges for practical deployment.

7. Research Status and Development Trends

7.1. Research Status

The rise in smart applications (IoT, smart cities, healthcare, electronic devices) demands the use of a large number of sensors [110]. Sensors based on electrochemistry, optics, and bionics are practically applied in MFH substances for quality assessment and component detection. Sensor functionality has transitioned from “single detection” to “system integration,” allowing for technology fusion. Combining AI sensory technologies, high-precision sensors, and precision instruments enables comprehensive evaluation from appearance to chemistry [111,112]. Examples: e-eye with NIR for goji berry appearance and component analysis [113]; e-nose with GC-MS for odor digitization and substance differentiation [114]. Sensor applications have expanded from active component quantification (e.g., flavonoids, saponins) to various safety-related components (heavy metals, pesticide residues, mycotoxins, microbial contamination), covering the full “farm-to-table” control chain [115].

Despite the significant application potential of sensor technology in the MFH field, numerous challenges remain in translating experimental results into practical industrial applications [116], and these challenges are extensively discussed in Section 4, Section 5, Section 6 of this review. First, cost is a primary constraint for industrial adoption. AI-based sensory systems, which are discussed in Section 4, involve complex sensor arrays, high equipment costs, and stringent hardware and software requirements, and these factors result in substantial initial investment. High-precision sensors, which are discussed in Section 5, often rely on noble metal nanomaterials, specialty optical fibers, or complex fabrication processes—for example, the multi-step modification in Section 5.1 and the specialty fiber structures in Section 5.4—to enhance detection performance, and these factors lead to high costs that hinder adoption by small and medium-sized enterprises and grassroots regulatory bodies [117]. Second, insufficient stability and anti-interference capability limit the expansion of practical application scenarios. As noted in Section 4.2, e-nose feature extraction faces risks of high-dimensional feature redundancy and overfitting. The electrochemical sensors in Section 5.1 are prone to fouling by macromolecules such as polysaccharides and tannins in complex matrices, and this fouling results in signal decay. Sensors based on biological recognition elements such as antibodies and enzymes, which are discussed in Section 3.3, struggle to maintain their activity over extended periods in complex environments [118,119]. Third, the biocompatibility and environmental stability of certain novel sensing materials, such as quantum dots and metal–organic frameworks, during long-term use in complex MFH matrices remain to be validated, and this situation raises concerns about service life [119]. These intertwined challenges collectively constitute the core barriers to transitioning sensor technology from a “laboratory tool” to an “industrial empowerment platform.”

7.2. Development Trends

Addressing the urgent quality control demands in the MFH field is a key task. Future trends in sensor technology aim to systematically resolve the core challenges identified above [120].

First, the focus is transitioning from “material innovation” to “stability and reliability.” As discussed in previous sections, existing sensors commonly suffer from signal drift, poor environmental adaptability, and inactivation of recognition elements, and these issues are addressed in Section 3.3, Section 4.2, Section 5.1 and Section 5.3. Future research should focus on developing novel sensing materials that combine high stability, strong anti-interference capability, and reusability. Specific approaches are as follows. One approach is employing inert matrix encapsulation to isolate interference from complex matrices. Another approach is introducing internal reference calibration mechanisms for self-compensation of signal drift. A further approach is designing antifouling interfaces to suppress nonspecific adsorption. Another approach is exploring synthetic recognition elements such as nucleic acid aptamers and molecularly imprinted polymers as alternatives to traditional antibodies or enzymes. These approaches are intended to fundamentally enhance long-term reliability and batch-to-batch consistency in real production environments [120].

Second, the focus is evolving from “functional integration” to “intelligent decision-making.” While AI sensory technologies described in Section 4 have achieved multi-dimensional perception, critical bottlenecks remain. These bottlenecks include high-dimensional feature redundancy, limited model generalization, and the computational complexity associated with multi-source data fusion. An example of limited model generalization is the overfitting risk posed by 128 features with a small sample size in Section 4.2, and an example of computational complexity is noted in Section 4.1. Future efforts should deeply integrate machine learning with edge computing. This integration is intended to construct intelligent sensing systems capable of adaptive learning. Such systems enable real-time recognition of complex samples, online correction of signal drift, and early warning of quality changes. In this way, a closed loop from “data acquisition” to “intelligent judgment” can be achieved [121].

Third, the focus is shifting from “technological feasibility” to “economic viability.” Currently, most high-precision sensors face challenges in large-scale deployment due to complex fabrication processes and high material costs, with examples including the specialty fiber SPR sensors in Section 5.4 and multi-step modified electrochemical sensors in Section 5.1. Future development should prioritize low-cost, easily manufacturable, and scalable sensor technologies, such as paper-based sensors, biomass-derived carbon material sensors, and smartphone-coupled portable detection devices [122,123]. Concurrently, a tiered and differentiated sensor configuration strategy should be established for different application scenarios, including enterprise online monitoring, grassroots rapid testing, and laboratory confirmation. This strategy is intended to achieve an optimal balance between technical performance and economic cost while ensuring core detection capabilities.

Fourth, the focus is expanding from “single-point detection” to “full-chain integration.” Current sensor applications are largely confined to individual stages. These stages include raw material identification, process monitoring, component quantification, and safety screening. A coordinated integration across the entire value chain from cultivation to distribution is lacking. Future efforts should combine IoT with smart packaging technologies, such as the colorimetric sensors explored in Section 6.2.3, to construct a comprehensive digital monitoring network covering the entire “farm-to-table” continuum. This approach enables precise traceability and dynamic quality control throughout the full life cycle of MFH products [124].

8. Conclusions

Under the “Healthy China 2030” strategy and thriving health industry, MFH substances’ dual attributes make quality, safety, and control core industry demands. Traditional techniques, limited by instrument dependence and complexity, struggle with full-chain rapid detection. Sensor technology, with high sensitivity, real-time response, and portability, has become key support for MFH quality control.

This review systematically summarizes mainstream sensor types in the MFH field: AI sensory technologies (e-eye, e-nose, e-tongue), high-precision sensors (electrochemical, infrared, biosensors), and optical sensors (fluorescent, SPR). Based on unique principles, they achieve multi-dimensional detection from appearance to chemistry. Regarding applications, sensor technology comprehensively covers four core stages: raw material identification/quality control, processing monitoring, active component quantification, and safety screening. Through single-technology optimization and multi-technology integration, a comprehensive “farm-to-table” quality control system has been established, providing efficient solutions for authenticity verification, origin tracing, process optimization, component analysis, and contaminant detection.

Currently, sensor technology in this field has evolved from “single detection” to “system integration.” Cross-integration with nanotechnology, artificial intelligence, and the Internet of Things has yielded innovative advances. However, the translation of these technological breakthroughs into practical industrial applications still faces significant challenges. These challenges are particularly evident in terms of economic feasibility.

From an economic feasibility perspective, the following points can be observed. First, AI-based sensory systems, which are discussed in Section 4, require complex sensor arrays, high equipment costs, and stringent hardware and software requirements. These factors make such systems more suitable for large enterprises or third-party testing platforms rather than for small-scale operations. Second, high-precision sensors, which are discussed in Section 5, include specialty fiber SPR sensors and electrochemical sensors modified with noble metal nanomaterials. These sensors involve complex fabrication processes, high material costs, and demanding maintenance requirements. As a result, their use is currently limited to research and high-end testing scenarios. Third, biosensors, which are discussed in Section 3.3, have high specificity. However, they face economic burdens due to the high cost of biological recognition elements such as antibodies and enzymes, stringent storage conditions, and batch-to-batch variability. These factors complicate industrial quality control. Fourth, low-cost alternatives such as paper-based sensors and biomass-derived carbon materials show promise in research. Nevertheless, they have not yet reached industrial application standards in terms of detection sensitivity, long-term stability, and scalability. Their commercial maturity therefore remains in need of improvement.

Therefore, to promote large-scale application of sensor technology in the MFH field, economic feasibility must be incorporated as a core design criterion. This should be done while ensuring detection performance. Priority should be given to developing low-cost materials, simplified fabrication processes, and high reusability. Concurrently, a tiered and differentiated sensor configuration strategy should be established for different application scenarios. These scenarios include enterprise online monitoring, grassroots rapid testing, and laboratory confirmation. The goal is to achieve an optimal balance between technical performance and economic cost. Only in this way can sensor technology truly transform from a “laboratory tool” into an “industrial empowerment platform.” Such a transformation can provide sustainable technical support for the high-quality development of the MFH industry. It can also contribute more robustly to public health.