Modernizing Asthma Diagnostics: Biosensors Enhanced by Nanomaterials and Artificial Intelligence

Abstract

Asthma is a prevalent, long-term inflammatory airway condition that is difficult to diagnose and treat because there is no single reliable diagnostic test. Misdiagnosis is therefore common, with rates as high as 73% in juvenile groups and up to 35% in adult populations. This ultimately exacerbates their illness by postponing therapy for some people and administering needless medication to others. Although well-known biomarkers such as blood eosinophils and fractional exhaled nitric oxide, as well as conventional diagnostic techniques such as spirometry, have improved clinical assessment, they are nevertheless constrained in many healthcare settings by limited availability, high cost, and inconsistent use. Furthermore, these indicators primarily reflect type-2 inflammation and are less useful for non-type-2 asthma, highlighting the need for more comprehensive, readily accessible diagnostic techniques. Identifying novel biomarkers of oxidative stress, metabolic alterations, and airway inflammation, including volatile organic compounds and redox-related chemicals, has been the focus of recent studies. These biomarkers offer opportunities for improved disease phenotyping and non-invasive detection. Simultaneously, advances in biosensor technology have enabled highly sensitive platforms to rapidly detect these biomarkers at low concentrations. In particular, optical biosensors are becoming more and more popular due to their ability to do real-time detection without the need for labels and their ease of miniaturization for point-of-care devices. This work summarizes traditional diagnostic tools alongside existing information on asthma phenotypes and clinically important biomarkers, and discusses advanced biosensors ranging from electrochemical to optical systems, including recent developments in nanomaterial-enhanced optical biosensing techniques. The importance of artificial intelligence and smartphone-integrated hardware is also covered, along with the main challenges that need to be overcome for these technologies to become useful clinical tools for asthma diagnosis and monitoring.

1. Introduction

Asthma is a non-communicable chronic inflammatory airway disease impacting about 300 million people globally, with an estimated 461,000 fatalities recorded in 2019, underscoring its significant contribution to morbidity and mortality worldwide [1]. The lack of a definitive gold-standard test and the challenges of obtaining an accurate diagnosis due to non-specific symptoms exacerbate this problem. According to recent population-based research, about 20–73% of people with current asthma remain undetected, delaying medication and airway remodelling, while about 30–35% of adults are over-diagnosed and undergoing needless treatment. These errors raise healthcare expenses, expose patients to needless corticosteroids, and delay the diagnosis of other cardio-respiratory disorders, including coronary artery disease [2,3,4]. Similar challenges are observed in the pediatric population, where studies have reported that approximately 53.5% of children diagnosed with asthma are over-diagnosed, whereas 50–68% of true asthma cases remain undetected [5,6,7]. Due to poor cooperation, sporadic symptoms, and the heterogeneous nature of asthma, traditional diagnostic tools, such as spirometry paired with clinical history, frequently fail to accurately identify asthma, particularly in children [8]. Spirometry may be normal despite underlying asthma and requires patient cooperation, trained personnel, and specialized equipment, whereas peak expiratory flow, bronchoprovocation, and fractional exhaled nitric oxide (FeNO) testing are constrained by variable accuracy, biological and environmental variability, interpretation challenges, cost, and limited accessibility, which further delays proper diagnosis and exacerbates global health inequities [9,10,11]. These shortcomings highlight the pressing need for accessible, objective, and non-invasive point-of-care diagnostic techniques to enhance disease classification and early diagnosis. Asthma is increasingly viewed as a heterogeneous condition encompassing several phenotypes, including T2-high and T2-low endotypes, which are influenced by distinct inflammatory pathways. Particularly for biologic treatments targeting inflammatory pathways such as interleukin (IL)-5 and interleukin (IL)-4/IL-13, biomarkers including immunoglobulin E (IgE), FeNO, blood eosinophils and sputum eosinophils have enhanced phenotypic classification and therapeutic stratification. However, high prices, the requirement for specialized equipment and knowledge, variations in sample collection and examination techniques, and overlapping biological data limit the therapeutic use of many biomarkers [12,13]. Biosensors have been identified as a promising substitute to traditional diagnostic methods, enabling rapid, sensitive, on-site detection of disease biomarkers while reducing reliance on complex laboratory procedures, trained staff, and expensive analytical instruments [14]. By leveraging various sensing modalities, including electrochemical, piezoelectric, and optical platforms, recent technical developments now enable the objective detection of asthma-related liquid or gaseous biomarkers at parts-per-billion levels [15,16]. Among these, optical biosensors have drawn particular attention because of their label-free detection capability, high sensitivity, and compatibility with miniaturized and portable platforms. Transduction techniques such as Raman spectroscopy, surface plasmon resonance, and fluorescence quenching allow real-time monitoring of biomarkers, such as volatile organic compounds (VOCs), associated with airway inflammation [17,18]. Nanomaterials such as graphene, quantum dots, carbon nanotubes and metal nanoparticles have high surface area, better signal amplification, and improved optical properties, which are very useful for further improving the performance of these biosensors by lowering detection limits, increasing sensitivity, and helping achieve faster response times [19]. The use of these sensors with smartphones and artificial intelligence/machine learning (AI/ML) algorithms supports decentralized testing and allows automated analysis of obtained data, making mobile devices useful diagnostic tools that can assist in the precise identification of specific biomarkers and disease monitoring [20,21]. Although there is scientific literature on asthma biomarkers, biosensing platforms for lung diseases, AI-based asthma management, and nanoparticle theranostics for respiratory disorders, little has been done to integrate these areas into a coherent framework [22,23,24,25]. The current paper seeks to address this knowledge gap by providing an extensive review of asthma phenotypes and emerging biomarkers for the diagnosis and management of asthma, while highlighting the shortcomings of existing diagnostic modalities. Moreover, the paper discusses how various biosensors, such as electrochemical and optical systems, are transforming non-invasive diagnostics. By emphasising the integration of nanomaterials with optical platforms, smartphones, and AI, the review underscores the translational strategies required to design the next generation of asthma-monitoring devices.

2. Conventional Test for Asthma

The primary factors used to diagnose asthma are symptoms, including coughing, shortness of breath, episodic wheezing, and chest tightness, which indicate varying airflow obstruction. Objective testing, a physical examination, and a clinical history are required to confirm airflow limitation [26]. Standard diagnostic techniques used to identify airflow obstruction or airway hyperresponsiveness include spirometry, bronchodilator reversibility testing, peak expiratory flow variation, and bronchial provocation testing. Given the heterogeneous nature of asthma, diagnosis depends on a combination of clinical assessment and these tests to reduce the risk of misdiagnosis, as no single gold-standard test exists [27].

2.1. Initial Evaluation—History and Physical Examination

For all age groups, a careful clinical history and physical examination form the foundation of asthma diagnosis. Asthma symptoms, such as wheezing, dyspnea, and coughing, often change over time and may worsen at night or in response to stimuli such as allergens and exercise. While these symptoms provide important clues, their diagnostic value is limited because patient-reported symptoms can be inconsistent and nonspecific. Therefore, diagnosis requires physician observation, often supported by standardized questionnaires to confirm symptoms, their pattern, and probable triggers, such as exposure to dust, pollen, and smoke. The likelihood of asthma is further supported by the existence of allergic disorders such as eczema, allergic rhinitis, or food allergies, together with a family history of asthma [28,29]. A physical examination includes assessing vital signs and performing head, neck, and pulmonary assessments. While an examination may be normal in between episodes, patients may exhibit wheezing, hyperresonance, prolonged expiration, and hyperinflation. Tachycardia, tachypnea, pulsus paradoxus, and trouble speaking can occur during severe episodes. The diagnosis may also be supported by atopic features, such as allergic rhinitis, eczema, atopic dermatitis, nasal polyps, pale nasal mucosa, and pharyngeal cobblestoning [30,31,32].

2.2. Spirometry

Spirometry is the preferred objective test for suspected asthma as it accurately measures airflow limitation. It determines lung function by measuring forced expiratory volume in one second (FEV₁) and forced vital capacity (FVC). A decline in the FEV₁/FVC ratio, usually less than 0.70 in adults, indicates airflow obstruction. Improvements in lung function following bronchodilator usage, with an increase in FEV₁ of >12% and >200 mL, demonstrate reversibility [33]. Variability in airway obstruction is another characteristic of asthma, as shown by serial spirometry, which shows variations in FEV₁ over time, either an increase or a reduction of >12% and >200 mL from baseline. However, in those with suspected asthma, especially if there are no symptoms, spirometry may seem normal because of the heterogeneous nature of asthma and its variable, often sporadic airflow limitation. Over time, airway obstruction may change and resolve spontaneously, after treatment, or during well-controlled periods. As a result, despite underlying bronchial hyperresponsiveness and airway inflammation, normal spirometric readings may be observed. Therefore, asthma cannot be ruled out by normal spirometry, and more testing may still be required [34]. In addition, obtaining accurate spirometry requires compliant patients who can perform the technique correctly, well-trained operators, and properly maintained and calibrated equipment. Interpretation of spirometric values also requires experience and consideration of factors such as age, sex, height, and ethnicity, which can limit its practical use in some clinical settings [9]. Even though conventional spirometry is often used to evaluate lung function, its routine use in primary care settings may be limited by equipment, staffing, and clinical resource requirements. To address these limitations, home spirometry has been proposed as an alternative that enables patients to repeatedly measure lung function in their normal daily environment. There is limited information available concerning the use of home spirometry as a component of the standard primary care diagnostic process. Specifically, the accuracy, adherence to periodic testing, and detection of daily variations in lung function, which are commonly overlooked in a single-visit spirometry test, remain undetermined [35].

2.3. Bronchoprovocation Testing

Bronchial hyperresponsiveness (BHR) is an exaggerated airway constriction in response to a stimulus and is a key feature of asthma, though it can also occur in other diseases and even in healthy individuals. Bronchoprovocation tests are used to assess BHR, especially when asthma is suspected despite normal routine spirometry results. They are more helpful for excluding asthma than for confirming it [36]. These tests fall into two categories, selective (e.g., chemical sensitizers, allergens or aspirin) and non-selective, which include stimuli such as histamine, methacholine, exercise, or mannitol. Non-selective stimuli are further divided into direct and indirect challenges. Direct bronchial challenge tests, such as methacholine and histamine, induce bronchoconstriction by acting on airway effector cells, like smooth muscle cells, mucus-producing cells and vascular endothelial cells, resulting in airway narrowing. These tests are highly sensitive in identifying airway hyperresponsiveness, but their specificity for asthma is limited, as positive responses may also occur in healthy individuals and patients with other respiratory conditions. Consequently, direct challenges are primarily used to rule out asthma when test results are negative [37,38]. In contrast, indirect challenges, including exercise, hypertonic saline, eucapnic voluntary hyperpnea (EVH), cold air, mannitol, and adenosine monophosphate (AMP), provoke bronchoconstriction indirectly by increasing airway surface osmolarity and stimulating inflammatory cells and neuronal cells to release mediators such as histamine, prostaglandins, neuropeptides and leukotrienes [39]. These mediators subsequently act on airway smooth muscle receptors, resulting in bronchoconstriction and a transient drop in FEV₁. Indirect tests are often more specific for identifying active asthma and are especially helpful for diagnosing exercise-induced bronchoconstriction because they more precisely represent the underlying airway inflammation. However, routine clinical use of many indirect challenge tests is limited by their time-consuming nature and need for specialized equipment [40]. Airway responses are evaluated by measuring the fall in FEV₁, and results are expressed as the dose required to cause this fall, known as PC20/PD20 (provocative concentration/dose causing a 20% fall in FEV₁). Bronchoprovocation testing has contributed significantly to understanding the pathophysiology of asthma and the relationship between airway inflammation and bronchial responsiveness. However, BHR is nonspecific, variable, overlaps between healthy and diseased individuals, and may persist despite treatment, making interpretation difficult [41].

2.4. Peak Expiratory Flow Monitoring

Peak expiratory flow (PEF) monitoring is one of the most common practices in diagnosing and treating asthma because it provides a quick and easy way to measure airway function. PEF reflects the extent of airway narrowing and represents the maximum airflow achieved after a forceful exhale following full inspiration; however, because it is a highly effort-dependent measure, poor patient cooperation, incorrect technique, or coughing during the maneuver can affect its accuracy. The measured value is also influenced by several physiological and demographic parameters. Demographic factors such as height, age, and sex affect lung size and airway calibre, while physiological factors, including lung elastic recoil, airway resistance, chest wall mechanics, and respiratory muscle strength, further determine airflow, particularly during bronchospasm or airway inflammation. Measurement of PEF is commonly performed using peak expiratory flow meters, which are easy to use, inexpensive, and suitable for home use as well as for use in primary care and emergency settings where spirometry may not be available [42,43]. Serial measurements of peak expiratory flow rate (PEFR) over 1 to 2 weeks can be used to evaluate airflow obstruction and provide a simple and practical approach for both asthma diagnosis and follow-up. Serial PEFR recordings may also show reversible airflow limitation through diurnal fluctuations, a fall following exercise, daily or visit-to-visit fluctuation or improvement after use of a fast-acting bronchodilator. Since variability in airway obstruction is a key feature of asthma, a day-to-day PEF variability greater than 10% supports the diagnosis [33]. For self-management, PEF values are commonly incorporated into asthma action plans using a three-zone traffic-light system based on the patient’s personal best PEF value, where the green zone (80–100%) indicates good asthma control, the yellow zone (50–80%) signals worsening symptoms and the need for treatment adjustment, and the red zone (<50%) indicates severe airflow limitation requiring urgent medical attention. A peak expiratory flow below 200 L/min is generally considered indicative of severe airflow obstruction in most adults under 65 years of age [44]. However, the utility of PEF monitoring in the long run is limited by its dependence on patients’ effort and proper technique, poor adherence to regular monitoring, device variability, and concerns about the accuracy of some home-based devices. In addition, PEF values are influenced by factors such as age, sex, and height and are less sensitive than spirometry for identifying mild airflow obstruction and small-airway dysfunction; therefore, results should be interpreted carefully and used as an adjunct rather than a replacement for spirometry [11,28,45]. These conventional methods are useful for diagnosing asthma, but they have significant disadvantages that highlight the need for biomarker-based approaches that can increase precision and help identify disease subtypes and guide treatment [46].

3. Asthma Phenotypes and Emerging Biomarkers

Asthma is a multifaceted disease having a broad range of clinical manifestations, severity, natural history, underlying pathophysiological causes, and treatment responses. Although asthma is typically classified into allergic (extrinsic) and non-allergic (intrinsic) forms, contemporary methods such as supervised clinical classification, unsupervised data-driven clustering, and systems biology have increased our understanding of asthma heterogeneity and allowed for the identification of unique phenotypes based on physiological, clinical, and biological traits [47,48]. Asthma phenotypes can be categorized based on clinical features such as early- or late-onset, obesity-related, smoking-associated, cough-variant, and exacerbation-prone asthma, as well as by etiology, such as allergic, non-allergic, aspirin-exacerbated respiratory disease (AERD), exercise-induced, and occupational asthma. Other classifications are based on pulmonary function patterns (reversible airflow limitation, airway hyperresponsiveness and fixed airway obstruction), treatment response or disease severity (mild, severe, steroid-resistant, and difficult-to-treat asthma), and airway inflammatory profiles (eosinophilic, neutrophilic, paucigranulocytic and mixed granulocytic) [49]. Collectively, asthma phenotypes represent the clinically observable expression of the disease, combining diverse characteristics that do not necessarily directly relate to the underlying molecular or immunological mechanisms driving the condition. In contrast, asthma endotypes are biologically and mechanistically distinct subtypes of the disease, defined by specific underlying molecular and immunological pathways that drive airway inflammation and remodelling. Thus, while phenotypes describe how the disease appears clinically, endotypes explain what is causing the disease at the molecular level. These mechanisms allow for the general classification of asthma into two primary endotypes, Type-2 and non-Type-2, as shown in Figure 1, which can be further described using different biomarkers [50].

3.1. Type-2 (T2-High) Eosinophilic Asthma Biomarkers

T2-high asthma is a common inflammatory endotype characterized by eosinophilic airway inflammation caused by T helper 2 (T_H2)-mediated immune responses and associated cytokines, including IL-4, IL-5, and IL-13. The pathophysiological process is initiated by exposure to allergens, pollutants and pathogens, which activate pattern recognition receptors (PRRs) on airway epithelial cells. This activation causes epithelial damage and the release of alarmin cytokines, including thymic stromal lymphopoietin (TSLP), IL-25, and IL-33, as well as chemokines such as CCL2 (C–C motif chemokine ligand 2) and CCL20 (C–C motif chemokine ligand 20). These mediators are very important for triggering and amplifying type 2 immune responses by activating dendritic cells (DCs), type 2 innate lymphoid cells (ILC2s), and promoting T_H2 cell differentiation [51,52,53]. Activated dendritic cells promote naïve CD4⁺ T-cell polarization toward a T_H2 phenotype, while ILC2s help mediate the early, antigen-independent production of type 2 cytokines and thus maintain inflammation. The resulting cytokine environment is dominated by IL-4, which promotes T_H2 differentiation and IgE class switching in B cells, IL-5, which regulates eosinophil maturation in the bone marrow and prolongs their survival in peripheral blood and airway tissues and IL-13, which contributes to goblet cell metaplasia, mucus hypersecretion, airway hyperresponsiveness, and structural airway remodelling [54,55]. These pathways promote recruitment of eosinophils into the airways via adhesion and transmigration involving P-selectin glycoprotein ligand-1 (PSGL-1), vascular cell adhesion molecule-1 (VCAM-1), and integrins. Within airway tissue, eosinophils become activated and degranulate, releasing cytotoxic proteins such as eosinophil peroxidase and major basic protein, thereby perpetuating type 2 inflammation by causing epithelial injury, airway hyperresponsiveness, and chronic inflammation. Both allergic and non-allergic asthma fall under this endotype, which is frequently associated with eosinophilia and elevated IgE levels [56]. A convenient and less invasive biomarker of eosinophilic inflammation is the blood eosinophil count (B-EOC). Eosinophilia is classified into lower (<1500 cells/µL), moderate (1500–5000 cells/µL), or high (>5000 cells/µL) categories and is usually considered when the count exceeds 500 cells/µL [57]. Inconsistencies were reported in the diagnostic performance of blood eosinophils for asthma, with sensitivity ranging from 0.15 to 0.79 and specificity from 0.39 to 0.91, depending on cutoffs and populations. Even when thresholds are set near or above the upper limit of normal, accuracy would still be inadequate for use as a sole diagnostic tool, although specificity increases with higher cut-offs [58]. Several studies report a moderate correlation between blood eosinophils (B-EOS) and sputum eosinophils (S-EOS). For example, B-EOS >220 cells/µL predicts S-EOS ≥3% with 70% specificity and 77% sensitivity, whereas a cut-off of ≥0.27 × 10⁹/L yields 91% specificity and 78% sensitivity [59,60]. Despite having moderate overall diagnostic performance (AUC ≈ 0.78), B-EOC demonstrates poor reproducibility across studies, and its discriminative ability remains limited. It is affected by numerous clinical and demographic factors, including sex, age, race, smoking, obesity, metabolic syndrome, atopy, allergic rhinitis, and body mass index, and can vary widely even in stable asthma [61,62]. Studies conducted over time intervals have found considerable intra-individual variability: only approximately 22% of patients showed levels above 300 cells/µL for five years, making it unsuitable for stable phenotyping [62]. Moreover, eosinophils are pleiotropic and heterogeneous, with different phenotypes and activation states, making it difficult to rely on a single blood measurement to accurately assess airway inflammation. Therefore, a patient may have a low B-EOC level even with airway eosinophilia (false negatives), or a high B-EOC level even without sputum eosinophilia (false positives). In contrast, sputum eosinophil measurement provides a direct measure of airway inflammation and represents the scientific gold standard, even though it is less practical in everyday practice. Thus, while B-EOC is a convenient surrogate marker of eosinophilic airway inflammation, it cannot be regarded as a reliable substitute for sputum eosinophil assessment [46,61,63]. Fractional exhaled nitric oxide (FeNO) is a non-invasive biomarker used to assess airway inflammation by measuring the concentration of nitric oxide (NO) in exhaled breath. The production of NO in the respiratory tract is mediated by airway epithelial cells, neurons, vascular endothelial cells, and even immune cells such as neutrophils and macrophages. The synthesis involves the conversion of the amino acid L-arginine into nitric oxide through the action of nitric oxide synthase (NOS) enzymes, which include constitutive forms, neuronal NOS (nNOS) and endothelial NOS (eNOS), producing low levels of NO for airway homeostasis, and inducible NOS (iNOS), which is transcriptionally activated and produces large amounts of NO during inflammation. Under normal conditions, constitutive NO promotes bronchodilation via cyclic guanosine monophosphate (cGMP)-mediated smooth muscle relaxation and plays a role in normal airway defence and function [64,65]. In type 2 airway inflammation, cytokines, particularly IL-4 and IL-13, increase NO production by activating STAT6 (signal transducer and activator of transcription 6)-dependent pathways that increase iNOS expression in airway epithelial cells. This is further influenced by reduced L-arginine availability due to arginase inhibition and the action of asymmetric dimethylarginine (ADMA), which disrupts normal nitric oxide homeostasis. Excess NO can react with superoxide to form reactive nitrogen species such as peroxynitrite, which contribute to airway damage and inflammation. The NO generated in the airway diffuses into exhaled breath, where it is measured as FeNO as an indirect marker of iNOS-mediated type 2 airway inflammation [66,67]. The American Thoracic Society recognizes elevated FeNO levels, above 50 ppb in adults and above 35 ppb in children, as a marker of substantial T2-mediated inflammation. Likewise, Global Initiative for Asthma (GINA) considers a threshold of >50 ppb to support a diagnosis of T2 asthma [68,69]. The diagnostic performance of FeNO varies depending on the chosen cutoff value. At a cutoff level of 25 ppb, the sensitivity and specificity of FeNO were 75.4% and 47.9%, respectively. Conversely, at a higher cutoff of >50 ppb, FeNO exhibited high specificity (99%) but low sensitivity (24%), suggesting FeNO is better for confirming asthma when values are high rather than ruling out the condition when results are low [70,71]. Serum periostin is a protein found in the extracellular matrix that indicates T2 inflammation and airway remodelling, and its production is induced by IL-4 and IL-13. According to a logistic regression analysis of 59 patients, the most reliable indicator of airway eosinophilia was serum periostin (p = 0.007) [51,72]. Immunoglobulin E (IgE) is another biomarker of eosinophilic inflammation and atopy that promotes allergen-induced asthma by activating mast cells and basophils and recruiting eosinophils. Its elevated level is linked to allergen sensitisation, airway hyperresponsiveness, and decreased lung function [73]. The IgE POCT (Niji™) point-of-care diagnostic tool has recently made it possible to detect IgE-mediated allergy reactions in-office by quantitatively measuring total IgE from finger-stick blood in around 12 min [74]. Integration of these biomarkers allows precision medicine in asthma by detecting specific inflammatory endotypes and predicting responsiveness to targeted biologics (anti-IL-5, anti-IL-4/IL-13 and anti-IgE). This approach improves treatment outcomes while lowering severe exacerbations and needless corticosteroid use by allowing biologics to be prescribed to patients most likely to benefit [75].

3.2. Non-Type-2 (T2-Low) Asthma

T2-low asthma is marked by the lack of classical T2-high markers, like eosinophilia, due to non-Type 2 immune responses which are distinct from the IL-4/IL-5/IL-13-driven response central to T2-high asthma, and is usually linked with paucigranulocytic or neutrophilic airway inflammation, along with reduced responsiveness to corticosteroids. Neutrophilic asthma is characterized by a preponderance of airway neutrophils, greater disease severity and exacerbations, and reduced bronchodilator responsiveness. Paucigranulocytic asthma, on the other hand, is defined by minimal airway infiltration by both eosinophils and neutrophils, yet patients still exhibit typical asthma features such as reversible airflow obstruction and airway hyperresponsiveness [76]. The T helper 1 (Th1) and T helper 17 (Th17) immune pathways are the primary drivers of T2-low asthma. Airway epithelial injury triggered by pathogens and irritants initiates the release of alarmins such as IL-33 and thymic stromal lymphopoietin (TSLP), which activate dendritic cells and promote the production of IL-23 and IL-6, thereby driving Th17 differentiation [77]. As a result, IL-17A produced by Th17 cells along with CD8⁺ T cells, γδ T cells, natural killer cells (NKT) cells, and type 3 ILCs (ILC3s) induces epithelial production of neutrophil-attracting chemokines such as CXCL1 (C–X–C motif chemokine ligand 1) and IL-8, leading to recruitment and accumulation of neutrophils in the airways, a key feature of this endotype. These neutrophils contribute to airway injury through proteases, oxidative stress, degranulation, and the formation of neutrophil extracellular traps (NETs), which further amplify epithelial damage and inflammatory signalling (including IL-1β, IL-6, and IL-8). IL-17 also contributes to structural airway changes by enhancing goblet cell overgrowth and stimulating proliferation of airway smooth muscle cells, thereby promoting airway remodelling. In addition, IL-17 and IL-23 contribute to corticosteroid resistance by suppressing glucocorticoid receptor (GR)-α activity while increasing expression of the inhibitory GRβ, a receptor isoform that blocks GRα function. As a result, the anti-inflammatory effects of corticosteroids become weak, contributing to steroid resistance [78,79,80]. Th1 immunity, driven by interferon (IFN)-γ–producing T cells and ILCs, shows a dual pattern in asthma. Some patients show reduced epithelial interferon responses, increasing susceptibility to viral infections and exacerbations, while others show excessive IFN-γ activity along with increased CXCL10 and IFN regulatory factor 5, which is linked to steroid resistance and more severe airway inflammation [77,80,81]. Clinically, T2-low asthma is often linked with phenotypes such as obesity-associated, very late-onset, and smoking-associated asthma, although its exact mechanisms and reliable biomarkers remain poorly understood [47]. Unlike T2-high asthma, it lacks a validated diagnostic biomarker and is typically identified by low levels of classical T2 markers (blood eosinophils, FeNO, and IgE) together with neutrophilic or paucigranulocytic airway inflammation. Elevated sputum neutrophil counts remain one of the most widely used markers of neutrophilic airway inflammation; while blood neutrophil counts may provide supportive evidence, their association is less consistent [82]. Several other biomarkers also reflect neutrophilic inflammation, including IL-8 (CXCL8), a key mediator of neutrophil recruitment, as well as myeloperoxidase (MPO) and neutrophil elastase that indicate ongoing airway injury [81]. Elevated levels of IL-17 and IL-1β have also been linked to neutrophilic inflammation, disease severity, and corticosteroid resistance through their roles in Th17-driven immune responses [83]. In addition, new findings have identified microRNAs (miRNAs) as potential biomarkers because they regulate gene expression and exhibit altered patterns in immune cells, airway epithelium and smooth muscle cells. Some miRNAs are also associated with steroid resistance and impaired lung function in neutrophilic asthma [49]. Other candidate markers include tumor necrosis factor (TNF)-α, YKL-40 (Chitinase 3-like protein 1), S100 calcium-binding protein A9 (S100A9), serum amyloid A1 (SAA1), folliculin and NET-associated products (e.g., extracellular DNA) [77]. However, none of these have yet been clinically validated as non-invasive biomarkers, underscoring a major research gap in the diagnosis and treatment of T2-low asthma.

3.3. Breath Volatile Organic Compounds (VOCs)

Volatile organic compounds (VOCs) in exhaled breath are promising non-invasive biomarkers for asthma because they represent airway inflammation, oxidative stress, and metabolic alterations associated with the disease. These compounds arise from both external environmental exposures and endogenous mechanisms such as inflammatory cell activation, lipid peroxidation of polyunsaturated fatty acids and airway microbiota metabolism [84,85]. During airway inflammation, activated inflammatory cells such as eosinophils, neutrophils, and macrophages produce excessive reactive oxygen species (ROS), including superoxide radicals, hydroxyl radicals, and hydrogen peroxide. These ROS attack polyunsaturated fatty acids present in cell membrane phospholipids, initiating lipid peroxidation. During this process, lipid radicals react with oxygen to form lipid peroxyl radicals and lipid hydroperoxides, which subsequently decompose into secondary oxidation products such as malondialdehyde (MDA), isoprostanes, and 4-hydroxy-2-nonenal (4-HNE). Further breakdown of oxidized lipids generates volatile compounds that can be detected in exhaled breath [86,87]. These VOCs primarily belong to five major categories that include hydrocarbons (alkanes, e.g., ethane and pentane), aldehydes (e.g., hexanal, nonanal), ketones, alcohols, and aromatic compounds. The potential of breath VOC profiling for diagnosis, phenotyping, and monitoring is highlighted by its ability to differentiate eosinophilic and neutrophilic phenotypes, identify atopic versus non-atopic asthma, and predict corticosteroid responsiveness, disease control, and exacerbations [85,88]. Platforms such as electronic noses (e-Noses), gas chromatography (GC), ion mobility spectrometry (IMS), mass spectrometry (MS) and hybrid platforms such as GC–MS and GC–IMS are frequently used to analyze VOCs. While chromatographic and spectrometric techniques separate and identify individual compounds for qualitative and quantitative analysis, sensor-array systems like the e-Nose detect overall breath fingerprints [89,90]. A systematic review and meta-analysis reported that exhaled VOC profiling has high diagnostic accuracy for asthma, with sensitivities, specificities, and area under the curve (AUC) of 87%, 86%, and 0.94, respectively [88]. Furthermore, it has been demonstrated that a portable gas chromatography system can do real-time breath VOC analysis in around 30 min, facilitating the rapid detection of biomarkers linked to asthma [91]. However, contributions from exogenous factors complicate the interpretation of exhaled VOCs, as these compounds may arise from environmental pollutants, smoking, diet, medications, physical activity, or microbial metabolism in the gut and airways rather than solely from endogenous disease processes. VOCs present in the surrounding air or introduced during sample collection may also appear in exhaled breath, making it challenging to determine whether they originate from the disease process or external sources [88,92]. To reduce such interference, ambient VOC levels are measured and subtracted from exhaled concentrations using the alveolar gradient approach, while inhalation of VOC-filtered air before sampling and standardization of pre-test conditions, including avoidance of smoking, food, beverages, and exercise, are commonly employed to improve the reliability of breath VOC analysis [92].

3.4. Hydrogen Sulfide (H₂S) and Other Redox Biomarkers

Hydrogen sulfide (H₂S) has gained recognition as a potential biomarker in asthma due to its roles in airway inflammation, oxidative stress, and redox regulation. Enzymes like cystathionine-β-synthase (CBS), cystathionine-γ-lyase (CSE) and 3-mercaptopyruvate sulfurtransferase (3-MST) in tissues like primary lung fibroblasts and airway smooth muscle cells produce endogenous H₂S in the circulation, which is thought to function as an antioxidant and anti-inflammatory signalling molecule that controls reactive oxygen species and inflammatory pathways [93]. According to clinical studies, serum H₂S levels are generally lower in people with asthma, with adults exhibiting lower levels in stable asthma (55.8 ± 13.6 μM) compared to healthy controls (75.2 ± 13.0 μM), which further decreases during exacerbations, while asthmatic children also showed reduced levels (44.2 ± 11.0 μM) that correlated with lung function indices [94,95]. On the other hand, sputum H₂S levels are higher in asthmatic patients and are negatively associated with predicted FEV₁% and positively correlated with sputum neutrophil counts, indicating a neutrophilic airway [96]. Dysregulation of the redox system contributes to altered H₂S levels across asthma severity, with recent clinical studies showing a biphasic pattern: individuals with severe asthma exhibit lower exhaled H₂S levels (149.3 ± 70.7 ppb) than those with mild-to-moderate asthma (192.4 ± 57.1 ppb). [97]. Elevated H₂S in the early stage likely reflects a compensatory antioxidant, anti-inflammatory, and bronchodilatory response to airway inflammation. However, as asthma progresses, persistent oxidative stress, chronic inflammation, and airway remodelling may impair the activity of H₂S-producing enzymes, resulting in reduced H₂S production and loss of this protective mechanism, thereby contributing to progressive airway damage [98,99]. Differences in H₂S concentrations have also been reported between biological compartments. Sputum H₂S correlates with sputum neutrophils and IL-8 concentrations and primarily reflects local airway production and is therefore more closely linked to airway inflammation, whereas serum H₂S levels exhibit distinct changes during stable and exacerbated disease states. The opposite patterns observed in sputum and serum indicate that these compartments may reflect different aspects of H₂S regulation, with sputum H₂S appearing to be a more direct indicator of airway inflammatory processes. Consequently, the sputum-to-serum H₂S ratio has been proposed as a potentially useful biomarker for identifying ongoing asthma exacerbations and predicting future exacerbation risk [93,96]. Apart from H₂S, several redox biomarkers indicative of increased oxidative and nitrative stress are associated with asthma. These include lipid peroxidation markers (8-isoprostane, malondialdehyde) and protein modification products (3-bromotyrosine, chlorotyrosine) [100]. A range of established and emerging biomarkers, including their diagnostic accuracy, clinical applications, sample sources, and applicability to T2-low asthma, are summarized in

Table 1. Comparison of asthma biomarkers according to diagnostic performance, sample source, clinical application, and utility in T2-low asthma.

S. No	Biomarkers	Diagnostic Sensitivity (%)	Diagnostic Specificity (%)	Clinical Use	Sample Type	T2-Low Utility	References
1	Blood eosinophils (≥150–300 cells/μL)	~60–79%	~70–90%	Surrogate marker of eosinophilic airway inflammation, predicts exacerbations and response to corticosteroids and anti-IL-5 biologic therapy	Peripheral blood	Limited (typically <150 cells/μL)	[63,101,102]
2	Sputum eosinophils (≥3%)	~40–72%	~80–82%	The gold standard biomarker for airway eosinophilia and asthma severity stratification predicts corticosteroid responsiveness	Induced sputum	Moderate (<2% eosinophils + low neutrophils define the T2-low paucigranulocytic phenotype)	[103,104,105]
3	Fractional exhaled nitric oxide (FeNO) (>50 ppb in adults or >35 ppb in children)	~43–88%	~60–92%	Non-invasive marker of eosinophilic airway inflammation predicts inhaled corticosteroids (ICS) response	Exhaled breath	Limited (typically <25 ppb)	[46,102,106]
4	Total serum IgE (>150 IU/mL)	~33–89%	~20–82%	Identify allergen sensitization and T2 inflammation, determine eligibility and dosing of anti-IgE therapy (omalizumab)	Serum	Limited (Normal or low levels)	[46,107,108]
5	Serum periostin (~52 ng/mL)	~81.3–100%	~50–100%	Marker of IL-13-driven inflammation and airway remodelling/fibrosis, assesses disease severity, predicts anti-IL-13 response	Serum	Low (Generally elevated only in T2-high eosinophilic asthma)	[109,110,111]
6	Exhaled VOCs (breathomics	~75–91%	~86–100%	Promising non-invasive biomarker for asthma diagnosis and phenotyping via metabolic signatures.	Exhaled breath	High Potential (VOCs such as nonanal and hexane can identify T2-low/neutrophilic asthma endotypes)	[88,112,113,114]
7	Sputum Neutrophils	66.7%	73.3%	Identify neutrophilic asthma, often severe, associated with poor response to ICS, linked to airway remodelling, and may guide non-T2-targeted therapies (tezepelumab)	Induced sputum	High (identify T2-low asthma-neutrophilic asthma with >40% to >76% neutrophils and <2% eosinophils, and paucigranulocytic asthma with very low levels of both neutrophils and eosinophils).	[115,116]
8	Serum YKL-40 (Chitinase-3-like protein 1)	~100%	~98%	Biomarker of airway inflammation and remodelling, associated with asthma severity, exacerbations, and lower FEV1	Serum	High (Elevated levels are observed in neutrophilic and obesity-related asthma and are associated with T2-low inflammatory responses)	[117,118]

.

4. Biosensors

A biosensor is an analytical tool that detects a specific substance by combining a biological recognition element with a signal-generating system. Once the analyte interacts with a bioreceptor, such as an aptamer, enzyme, or antibody, a physicochemical phenomenon occurs, including changes in charge, colour, temperature, or weight. This change is then transformed into an electrical signal by a transducer. The signal is then processed by electronic components and presented in a user-friendly form [119,120]. Biosensors use different transducers and can be classified as electrochemical (measuring current, potential, or impedance), optical (using fluorescence or surface plasmon resonance), thermal (detecting reaction heat), and gravimetric/piezoelectric (tracking mass changes), each offering distinct benefits, from high sensitivity to affordability and scalability [121]. They are employed in many different fields, including health assessment, pharmaceutical research, monitoring of environmental contaminants or microorganisms, and detection of biomarkers [119].

Asthma Biosensors

Determining biomarkers associated with oxidative stress, immunological response, and airway inflammation, biosensors are crucial to the treatment of asthma. Electrochemical biosensors are particularly noteworthy among the various varieties because of their great sensitivity, real-time results, low sample requirements, and ease of integration into portable equipment. They function by converting biological interactions into electrical signals at biorecognition elements, typically enzymes. Depending on the function, additional components such as nucleic acids, antibodies, or even entire cells may be employed. These devices detect changes in impedance (impedimetric), voltage or potential (potentiometric), and current (amperometric); voltammetry and field-effect transistor (FET)-based sensing are more sophisticated techniques.

Electrochemical biosensors usually employ a three-electrode configuration comprising a reference, working, and counter electrode, and their performance depends on the material composition, surface area, and electron transfer rates. The use of nanomaterials like nanoparticles and nanowires makes it more sensitive and effective in terms of target capture because of a higher surface area to volume ratio [122,123]. More recently, metal single-atom (SA) nanoenzymes have been shown as excellent candidates for sensing materials because of their atomically dispersed active sites along with heteroatom coordination, which allows better electron transfer, high catalytic activity, and increased interaction with analytes. In contrast to conventional metal nanoparticles, the use of SA nanoenzymes (e.g., Fe single atoms) allows for optimized metal atom usage and provides highly accessible catalytic sites, resulting in enhanced sensitivity, selectivity, and overall electrochemical sensing performance [124]. Based on the principles of electrochemical biosensing, recently developed biosensors have demonstrated the ability to detect biomarkers associated with asthma. For example, the HexaPie sensor uses a label-free, non-faradaic impedance method to detect IL-8, IL-10, and IP-10 (interferon gamma–induced protein 10) directly in saliva. Limitations of detection (LOD) of 1.17 ng/mL (IL-8), 12.5 pg/mL (IL-10), and 0.20 ng/mL (IP-10) were achieved, with very little interference from components present in saliva. Its validation using ELISA and Luminex xMAP immunoassays confirmed a strong correlation for both IL-8 and IP-10, emphasizing its reliability as a small, reagent-free, point-of-care device for non-invasive immune analysis of asthma [125]. In another study, the researchers demonstrated the potential of a Ti₃C₂T_x MXene-based biofuel cell self-powered sensor (M-BSPS) for detecting the asthma marker MUC1 in saliva. This sensor has been fabricated with a CoNi-LDH/MXene anode and an Ag-NWs/MXene cathode, providing efficient ion diffusion, electron transfer, and charge storage for accurate electrochemical detection of target biomarkers. With a LOD of 65 pg/mL and a detection range of 0.1–750 ng/mL, the sensor showed good stability, selectivity, and reproducibility, proving its applicability for use in point-of-care asthma monitoring [126]. However, the long-term reliability of MXene-based sensors is a concern since oxidation and layer-restacking of the MXene can lead to a reduced active surface area, affecting the sensing performance. Thus, it is crucial to design oxidation-resistant MXene structures and develop cost-effective, scalable fabrication methods for their eventual clinical application in asthma monitoring [127].

Chemiresistive biosensors measure changes in the electrical resistance of the sensing material when a target binds to a sensing layer between interdigitated electrodes. They are simple, low-power devices that work well for wearable and point-of-care testing. Still, many of the conventional systems struggle with selectivity and slow response because they rely mostly on basic adsorption and weak or missing biorecognition elements. These limitations can be improved by using enzyme-functionalized nanomaterials along with light activation. When exposed to UV or visible light, the material produces charge carriers that boost conductivity and speed up the sensing process. ZnO nanorods are commonly used since they offer a large surface area, good biocompatibility, and enzyme immobilization, which helps improve electron transfer. Although this approach has been shown for lactate and glucose, the same idea could be used for asthma monitoring through biomarkers in exhaled breath [128,129]. A miRNA-21 sensing platform has been developed using a chemiresistive biosensor consisting of bilayer graphene (BLGR) interdigitated electrodes grown through direct chemical vapor deposition (CVD). It works by measuring the change in resistance due to hybridization of miRNA-21 with immobilized probe DNA, leading to decreased charge carriers’ mobility within the graphene channel. This improved response is mainly linked with increased interlayer coupling in directly grown BLGR compared with stacked graphene structures. Although the sensor shows high sensitivity and a low detection limit (5.2 fM), the manufacturing process includes multiple stages, including CVD growth, plasma treatment for surface modification, and fairly long DNA immobilization and hybridization times [130]. These steps make the overall process more complex and may need to be considered when thinking about reproducibility or scaling up for practical use.

Field-effect transistor (FET)-based biosensors use a semiconductor channel whose electrical behaviour changes when biological molecules bind to receptors on its surface, altering charge density and carrier mobility. This interaction affects signals such as drain current, threshold voltage, and conductivity, enabling real-time, label-free detection of biomolecules, including proteins, DNA, and cytokines. They are potential tools for early, non-invasive disease diagnosis because of their ability to detect low-level biomarkers even in complex samples like saliva and breath, along with their miniaturized and integrable design, making them well-suited for portable and lab-on-a-chip diagnostic applications [15,131]. However, FET-based biosensors face several challenges. In physiological fluids, the Debye screening effect can reduce the electrical signal generated by biomarker binding because surrounding ions partially shield the target charges from the sensor surface. In addition, variations in surface functionalization may affect reproducibility and long-term stability, limiting clinical translation [132]. In a study by Zhuang Hao et al., a portable graphene-based FET nanosensing system was developed to detect the cytokine biomarker interleukin-6 (IL-6) in saliva [133]. The device used an aptamer-functionalized graphene channel, in which IL-6 binding altered surface charge, resulting in measurable changes in electrical current. The technology responded rapidly within 400 s and achieved a detection limit of 12 pM, suggesting its potential for early, non-invasive monitoring of asthma-related inflammation. However, the authors noted signal variations among saliva samples from different individuals, which indicates that non-target components present in real biological samples can compromise sensing performance. Therefore, before the widespread routine usage in clinical practice, it is necessary to further improve sensor stability and carry out larger clinical validation.

Photoelectrochemical (PEC) biosensors are new-generation analytical platforms in which optical excitation, along with electrochemical sensing, has been applied to achieve high sensitivity and selectivity of biomarker detection through photocurrent production in photoactive materials. Nanomaterials, including quantum dots, graphene derivatives, metal–organic frameworks (MOFs), and semiconductor heterojunctions, have played a critical role in improving the PEC system performance via increasing light absorption, charge separation, and signal amplification. Multiplexed PEC biosensors reduce the amount of sample needed and analysis time while concurrently detecting different biomarkers. Their low background noise, rapid response, and compatibility with portable devices further add to their suitability for point-of-care applications [134]. In addition, the capability to multiplex detection of regulatory microRNAs, which appear to be very promising as markers of asthma-related airway inflammation and immunological disorders, has been illustrated by PEC biosensors. Considering the heterogeneity of asthma and involvement of multiple inflammatory pathways, such an approach may help in a more comprehensive estimation of disease phenotype, severity, and treatment response, highlighting that PEC biosensors offer considerable potential for asthma diagnostics and disease monitoring [135,136].

5. Optical Biosensors for Asthma

Optical biosensors are platforms that use light-based phenomena to detect and measure analyte concentrations in a sample. These biosensors detect interactions between two biological molecules, such as antigen–antibody binding or ligand-receptor interactions, to generate optical signals that can be measured and evaluated. Optical signals are observed as changes in physical properties, such as light intensity, wavelength, fluorescence, and the refractive index of the medium through which light passes. Such biosensors offer fast analysis, high sensitivity, and minimal interference from external electromagnetic radiation. For this reason, they are widely used in modern analytical systems and clinical diagnostics, including the detection of food toxins, environmental pollutants, and disease markers. Also, their compatibility with miniaturized platforms helps in the development of compact and user-friendly diagnostic devices [137,138]. Principles and applications of colorimetric, fluorescence, SPR (surface plasmon resonance), and SERS (surface-enhanced Raman scattering)-based biosensors for the detection and monitoring of asthma have been presented in Figure 2.

5.1. Colorimetric Biosensors

Colorimetric biosensors can be used for detecting a variety of analytes, including metal ions, biomolecules, nucleic acids, microorganisms, and clinical biomarkers. These biosensors operate based on the interaction of a particular target analyte with the sensing material that causes a visible color change, which makes detection fast and easy without necessarily using complicated equipment [139]. This color change occurs as a result of optical changes in metal nanoparticles (e.g., gold and silver) by localized surface plasmon resonance (LSPR), wherein the absorption properties of the particles change due to factors like nanoparticle aggregation or variations in the refractive index of the surrounding medium. In addition, colorimetric signals may also arise from chemical reactions, such as oxidation-reduction, enzymatic reactions, pH changes, or interactions with other ions and biomolecules. These sensors have been widely used for point-of-care and on-site analysis due to the ease in detecting the results visually, low cost, and fast responses [140,141]. A colorimetric biosensor based on enzyme-functionalized nylon–poly (allylamine hydrochloride) (PAH) nanofiber mats has been developed for the non-invasive detection of lactate, a potential biomarker of airway inflammation in asthma. The sensor captures lactate from breath aerosols and employs a lactate oxidase (LOx)–horseradish peroxidase (HRP)/TMB (3,3′,5,5′-Tetramethylbenzidine) assay to generate a visible color change. The platform showed detection limits of 5 μmol L⁻¹ (solution) and 20 μmol L⁻¹ (hydrogel), with a working range of 5–150 μmol L⁻¹ covering physiologically relevant breath lactate levels. Its simple visual readout and portability make it suitable for point-of-care monitoring of airway inflammation [142]. Despite the advantages, there have been a few reports of colorimetric biosensors which use asthma markers as targets. Although the biosensors are favourable for use at point-of-care due to their fast visual read-out without requiring any sophisticated instrumentation, their sensitivity may be inadequate in detecting asthma markers that are present in extremely low concentrations. In addition, the complexity of the biological samples can influence colour development, which affects the analytical accuracy and reproducibility, while interference from non-target species may cause false-positive results [143]. Although the described lactate sensor showed positive results, routine clinical adoption will not be possible unless sensitivity, stability, selectivity, and clinical validation are further improved. Nevertheless, the broader field of colorimetric respiratory sensing continues to progress. For example, dye-based platforms for monitoring exhaled CO₂ and multi-dye systems for detecting H₂O₂ in breath have demonstrated sensitive visual detection capabilities [144,145]. However, it should be noted that these devices cannot be considered true biosensors since they do not use specific biorecognition units, but they may inspire the development of colorimetric biosensors aimed at detecting asthma-specific biomarkers.

5.2. Fluorescence-Based Biosensors

Fluorescence-based biosensors work on the principle that certain molecules can absorb light at one wavelength and emit light at a longer wavelength, producing fluorescence. This technology combines the advantages of high sensitivity offered by the fluorescence technique with the selectivity provided by the molecular recognition element of fluorescence-based biosensors, which may consist of antibodies, ligand-binding proteins, or aptamers, to identify specific analytes. Various types of fluorescent probes, such as dyes, fluorescent proteins, and nanoparticles, such as quantum dots, can be used to improve detection outcomes. Fluorescence-based biosensors use different techniques, such as measuring fluorescence intensity, Förster resonance energy transfer (FRET), fluorescence polarization, and fluorescence correlation spectroscopy, to study molecular interactions, binding events, and diffusion processes [146,147]. Near-infrared (NIR) fluorescent probes can also be used to achieve improved sensing because NIR light can easily penetrate biological tissues and causes less background interference [148]. Overall, these biosensors are very sensitive, rapid, and affordable and have a wide range of applications in the detection of food contaminants, monitoring environments, and medical diagnostics [149]. An optical fibre-based multiplex biosensor with fluorescence detection has been reported to simultaneously detect the pro-inflammatory cytokines IL-6, IL-1β, and TNF-α via antibody sandwich immunoassay and fluorescence magnetic bead detection. This biosensor showed a detection range of 12.5–200 pg mL⁻¹ with a LOD of 12.5 pg mL⁻¹ for each cytokine. It has also exhibited excellent selectivity and performance similar to that of conventional ELISA. With the ability to detect multiple biomarkers at one time, this method may be able to offer a more complete analysis of airway inflammation compared to assays that measure a single biomarker, considering the complex cytokine networks involved in asthma pathogenesis. Though it can accommodate multiplexing, the assay involves many antibody-labelling and incubation steps and has mainly been validated in cell-culture samples [150,151]. In comparison with the multiplex optical fibre system, an aptasensor using FRET based on gold nanoparticles and nitrogen-doped carbon quantum dots achieved a significantly lower LOD of 0.82 pg mL⁻¹ for IL-6 detection, which reflects its high sensitivity. However, unlike the multiplex biosensor, it was limited to a single biomarker and showed a relatively narrow detection range [152]. To detect allergen-specific IgE involved in allergic asthma rapidly, a miniature biosensor based on fluorescence technology has been developed. This biosensor utilizes fluorescence-tagged antibodies to produce quantitative signals from the binding of IgE and allergens on microarrays, allowing multiplexed detection. With a portable fluorescence reader, it can analyze multiple allergens from a small serum sample (~80 μL) in about an hour, offering a compact and efficient approach for point-of-care identification of asthma-related allergic triggers [153]. Chai-Gao et al. also developed a fluorescence-based biosensor for multiplexed detection of allergen-specific IgE using a microfluidic platform and portable reader for rapid point-of-care analysis [154]. In these platforms, the portable readers used were less sensitive than commercial microarrays, and the validation process involved only a small number of control samples, highlighting the need for larger clinical studies to validate the tests further.

5.3. Surface Plasmon Resonance (SPR)-Based Biosensors

SPR-based optical biosensors are often used for label-free detection of biomolecular interactions. The method is usually applied in a Kretschmann configuration, in which a glass prism is coated with a thin layer of metal, usually gold, and exposed to monochromatic light from an LED or other light source. The SPR phenomenon occurs when polarized light strikes the thin metal surface at a specific angle at the interface of two media, such as glass and a liquid sample. At a specific angle, known as the resonance or SPR angle, this interaction reduces the intensity of the reflected light by transferring energy from incoming photons to free electrons on the metal surface, creating collective oscillations known as surface plasmons [17,155]. The principle of SPR is based on attenuated total reflection, where these plasmonic oscillations are highly sensitive to changes in the local dielectric environment. When biorecognition elements such as antibodies or enzymes are immobilized on the sensor surface and interact with target analytes, the resulting mass accumulation alters the refractive index at the interface. This results in a measurable shift in the SPR angle or resonance condition, which is detected using a photodiode array or similar detector system. The sensor creates a response curve, also known as a sensorgram, that depicts the interaction between molecules by tracking these changes over time [156]. A representative application of SPR-based biosensing in asthma involves detecting the inflammatory biomarker YKL-40 in patient serum. In a study by Naglot et al., YKL-40 levels were measured using a real-time SPR platform (BIAcore system), eliminating the need for labelling by immobilising specific antibodies onto a gold-coated sensor chip and tracking biomolecular binding via changes in refractive index [157]. With a detection limit of 0.33 ng/mL, the method demonstrated excellent sensitivity and enabled distinction between mild and severe asthma based on progressively rising YKL-40 concentrations. In addition to conventional single-marker SPR tests, a variety of SPRi-based arrays have been developed for highly sensitive biomarker multiplex analysis in complex biological samples. For instance, an SPRi biosensor for IL-6 in human plasma achieved highly sensitive, label-free analysis in the pg mL⁻¹ range (LOD ~ 0.3–0.9 pg mL⁻¹ depending on the recognition element). This platform was based on either antibody-functionalized or inhibitor-based receptors and had very good correlation with electrochemiluminescence assays (R² ~ 0.99), which means that results obtained with both techniques were almost identical across samples. It also provided very high selectivity against highly interfering substances in serum samples; however, the SPRi performance is critically dependent on surface chemistry and immobilization of receptors [158]. Another study used SPR-based biosensing to detect allergy-related asthma by monitoring both basophil activation and IgE-mediated responses in a label-free manner. This method uses FcεRI-expressing cells sensitised with patient serum to quantify IgE activity, while SPR imaging detects changes in refractive index resulting from basophil activation during allergen exposure. These combined strategies enable rapid, highly sensitive, label-free monitoring of IgE-driven basophil activation, but their clinical application is limited due to complex ex vivo processing of cells, surface-chemistry-dependent variability and relatively small-scale validation studies [159].

5.4. Surface-Enhanced Raman Spectroscopy (Sers)-Based Biosensors

Raman scattering occurs when incident light falling on molecules is scattered, with some of the scattered photons experiencing an energy shift due to molecular vibrations. This change in energy levels results in a Raman spectrum, which serves as a distinctive molecular identifier for the substance being analyzed. However, because only a small fraction of light undergoes inelastic scattering, the process generates a weak signal. To address this limitation, SERS has been developed as a highly sensitive biosensing technique. In SERS, molecules are adsorbed onto nanostructured metallic surfaces such as gold or silver, where localized surface plasmon resonance (LSPR) significantly strengthens the electromagnetic field and amplifies the Raman signal. This improvement makes it possible to detect biomolecules and volatile organic compounds at extremely low concentrations, allowing rapid and sensitive identification of specific analytes and disease-related biomarkers [160,161]. Although metal-based SERS substrates are highly sensitive, they usually lack uniform hotspot distribution, signal reproducibility, and dynamic tunability of optical properties after fabrication. Recent research is oriented towards designing novel nanostructured substrates which would allow improving the performance and reliability of SERS sensors [162]. Deng et al. proposed an innovative Ag/BaTiO₃ hollow microsphere array substrate, where plasmonic effects were complemented by pyroelectric field-induced charge transfer that resulted in improved signal uniformity and ultralow detection limits [163]. Even though this concept was proven using model compounds instead of biomarkers of asthma, the potential of dynamically tunable SERS platforms in future respiratory disease diagnostics seems promising. Nevertheless, the fabrication of such complex nanostructures remains challenging, and further studies are required to evaluate their long-term stability, scalability, and performance in real biological samples. In certain biosensing systems, antibody-modified gold nanorods are used to detect inflammatory cytokines, such as IL-5, in sputum samples. The nanorods assemble end-to-end upon IL-5 binding to the antibodies, forming nanoscale hot spots that significantly enhance Raman signal via SERS, enabling sensitive detection of IL-5 at extremely low concentrations. With a good correlation coefficient (R² = 0.994) and an enhancement factor of around 1.39 × 10¹, the biosensor showed a linear response in the 0.1–50 pg/mL range. The clinical study so far was conducted using a very limited number of patient samples, and there are insufficient comparisons with routinely used asthma diagnostic markers, which makes it difficult to fully establish its clinical applicability [164]. In comparison, the multiplex SERS–microfluidic immunoassay developed by Wang et al. represents an advancement in terms of analytical performance since it allows parallel measurement of several inflammatory interleukins (IL-6, IL-8, and IL-18) in complex blood plasma samples, rather than a single analyte identification [165]. The use of distinct Raman reporter molecules allowed each cytokine to be selectively encoded and quantified with low detection limits in the pg mL⁻¹ range and good reproducibility (RSD < ~8.5%), while also showing strong agreement with ELISA-based reference methods. Importantly, in contrast to the IL-5 nanorod system, this platform integrates a microfluidic architecture and chemometric (PCA) analysis, improving signal separation and robustness in multiplex biological detection. However, both approaches share common limitations, including strong dependence on surface chemistry, antibody immobilization, and non-specific binding control in complex samples. Moreover, although the multiplex SERS system provides more depth of analysis in terms of multi-biomarker identification, both systems are limited by small clinical data sets and require larger validation before routine asthma diagnosis. In addition, SERS has also been explored for breath-based diagnostics of respiratory diseases. For example, SERS platforms have been used to detect gaseous biomarkers such as nitric oxide in exhaled breath, which is associated with airway inflammation in asthma. In these devices, trace gas molecules in breath are captured by functionalized plasmonic substrates, and the resulting Raman spectral changes enable their sensitive detection even at extremely low concentrations. Additionally, wearable or breathalyzer-style SERS devices have been investigated for the analysis of breath aerosols and volatile biomarkers [166].

Collectively, these developments show that SPR and SERS biosensors have strong potential for rapid, sensitive, and non-invasive detection of disease biomarkers and real-time monitoring of respiratory disorders. However, their implementation for clinical purposes faces several limitations. SPR is relatively more established but is limited by its expensive instrumentation and system complexity, making it less suitable for routine or point-of-care use. Its performance depends greatly on external factors, including variations in refractive index, temperature and nonspecific adsorption affecting the reliability of signals in real samples. SPR biosensors also demonstrate relatively lower sensitivity than SERS. While SERS demonstrates greater sensitivity, the major challenge associated with it is reproducibility because of dependence on unpredictable hot spots and inconsistencies in nanostructure fabrication [167,168]. In complex biological samples, its application is subject to problems of biofouling, background interference, reduced signal-to-noise ratios, nanoparticle aggregation, coffee-ring effects, and non-uniform analyte distribution, thereby limiting measurement accuracy. In practice, in the case of SERS, sample dilution or preparation is necessary in order to avoid fouling issues, increasing the time and costs associated with analysis; moreover, anti-fouling approaches are hindered by their poor stability, incomplete coverage and insufficient reusability. As previously stated, in both cases, the lack of large-scale validation studies and standardized protocols is one of the main reasons why these methods have not been widely used yet in clinics, despite many years of research [169].

Selecting the optimal optical transduction mechanism among these emerging platforms is largely dependent upon the physical state and concentration of the asthma biomarker. Biomarkers that exist in gaseous form, such as FeNO and VOCs, are typically present in trace amounts and therefore can be detected using highly sensitive techniques such as SERS, which allow for molecular fingerprinting and detection of analytes at very low concentrations [170]. On the other hand, biomarkers in liquid-phase, including interleukins, YKL-40 and IgE, are commonly detected using SPR- and fluorescence-based biosensors. SPR is preferred for label-free, real-time detection of biomolecular interactions, whereas fluorescence-based biosensors are particularly useful for highly sensitive multiplex detection [150,155]. Comparison of optical biosensing platforms developed for asthma-related biomarker detection with an emphasis on biomarkers, sample types, principles of sensing, performance metrics, and other important aspects is summarised in

Table 2. Representative optical biosensors designed to detect asthma-related biomarkers in different biological samples.

S. No.	Optical Biosensor Type	Target Biomarker	Sample Type	Sensing Mechanism	Analytical Performance	Key Features	References
1.	Colorimetric biosensor (nanofiber mask platform)	Lactate (airway inflammation biomarker)	Exhaled breath aerosol	Nylon-PAH nanofibers capture lactate, followed by LOx/HRP–TMB enzymatic colorimetric reaction	LOD-5 μmol L−1 (solution), ~20 μmol L−1 (hydrogel) Range- 5–150 μmol L−1	Wearable face-mask platform enabling non-invasive breath biomarker monitoring	[142]
2.	Fluorescence-based biosensor (microfluidic microarray)	Allergen-specific IgE (allergic asthma biomarker)	Serum	Microfluidic allergen-functionalized micropillar array with fluorescence-labelled antibodies and optical reader	Detection sensitivity ~500 dye molecules/μm2 Spatial resolution ~50–100 μm Rapid readout (~1 s)	A portable fluorescence reader enabling multiplex detection of up to 88 allergens for allergy profiling	[154]
3.	Fluorescence-based aptasensor	VEGF165 (angiogenesis biomarker associated with bronchial asthma)	Serum	G-quadruplex aptamer–Thioflavin T fluorescence system, VEGF165 binding disrupts the G-quadruplex and reduces fluorescence intensity	LOD-0.138 nM Linear range-1.56–25 nM	Label-free aptamer-based fluorescent detection with good specificity and serum sample applicability	[171]
4.	SPR-based biosensor	Allergen-specific IgE	Serum	Gold SPR chip functionalized with 3-mercaptopropionic acid, followed by EDC/NHS coupling for immobilization of anti-IgE antibodies	Detection range-1–1000 ng/mL LOD-0.051 ng/mL LOQ-0.153 ng/mL	Label-free, highly selective detection with strong discrimination against BSA, IgG, and myoglobin	[172]
5.	SERS-based gas sensor	Hydrogen sulfide (H2S)	Exhaled breath	ZnO nanowire/Ag nanostructure coated with ZIF-8 metal–organic framework enriches H2S molecules and enhances Raman signal for detection	LOD-1 × 10−10 v/v RSD ≈ 7.13%	Flexible PVDF nanofiber membrane integrated into wearable SERS face mask for breath monitoring	[173]
6.	Electrochemiluminescence (ECL) biosensor	miRNA-221-5p (asthma-associated microRNA)	Saliva exosomes	Capture DNA hybridizes with miRNA-221-5p in saliva exosomes, generating an amplified ECL signal via Cu nanocluster@MXene.	LOD-34 aM Detection range-1.0 × 10−16 ∼1.0 × 10−8 M	Non-invasive saliva analysis with ultra-high sensitivity	[174]
7.	Electrochemiluminescence (ECL) biosensor	miRNA-126 (biomarker associated with asthma inflammation)	Saliva extracellular vesicles	Ti NC-SP ECL emitter on goldene interface with miRNA-126/DNA hybridization-based signal amplification.	Detection range-10−12–10 µM	Non-invasive saliva sampling, high sensitivity and signal stability, suitable for childhood asthma diagnosis	[175]

.

6. Next-Gen Biosensor Technologies

The progress made in recent years regarding biosensors has led to the emergence of next-generation platforms with increased sensitivity, rapid response time, and improved analytical capabilities for disease diagnosis. To improve signal transduction and facilitate the quick identification of low-abundance biomarkers, these technologies use innovations such as nanotechnology, microfluidics, artificial intelligence, wearable technology, and miniaturisation. As a result, next-generation biosensors are progressively being explored for non-invasive, real-time, and point-of-care monitoring of various diseases, including respiratory diseases [176].

6.1. Nanomaterial Enhancement of Optical Sensors

Nanomaterial integration has substantially accelerated the development of next-generation optical biosensors by improving sensitivity, selectivity, signal amplification, and overall analytical performance. Various nanomaterials, including metallic nanoparticles, quantum dots, and carbon nanomaterials, have been incorporated into optical sensing platforms to improve biomarker detection and support portable, real-time, and point-of-care diagnostic applications [177,178]. Apart from these, UCNPs have attracted considerable interest because their anti-Stokes emission enables near-infrared excitation with visible or ultraviolet emission, thereby reducing background autofluorescence and improving signal-to-noise ratios in biological samples. high photostability, tunable multicolour luminescence, and long emission lifetimes. The conversion of near-infrared excitation into visible or ultraviolet emission significantly reduces the effect of autofluorescence and improves signal-to-noise ratios in biological samples. However, the broader use of UCNPs in biosensing is limited due to low upconversion efficiency under mild excitation conditions, complicated synthesis methods, and the need for further studies on their long-term biocompatibility and safety [179,180]. The major types of nanomaterials used in optical biosensors and their role in sensing are illustrated in Figure 3.

6.1.1. One-Dimensional Nanomaterials

One-dimensional (1D) nanomaterials are nanoscale structures with one dimension significantly larger than the other two (1–100 nm). These materials, which include nanowires, nanotubes, and nanorods, have an anisotropic structure that allows tunable LSPR by varying their aspect ratio. This property supports strong optical absorption across the visible-to-near-infrared region, improving signal transduction in multicolorimetric and SERS sensing and increasing sensitivity for biomolecular detection [177]. Carbon nanotubes (CNTs), particularly single-walled carbon nanotubes (SWCNTs), have a quasi-one-dimensional cylindrical structure, a high surface-to-volume ratio, a large aspect ratio, and high chemical stability, which makes them ideal for optical biosensors. Because of these features, their optical properties are extremely sensitive to even minute changes in their surroundings.

Semiconducting SWCNTs provide highly sensitive optical detection through their inherent near-infrared (NIR) photoluminescence (650–1400 nm), which undergoes measurable shifts or quenching upon target binding. Multiplexed detection is made possible by their chirality-dependent properties, which also allow them to efficiently quench fluorescence via Förster resonance energy transfer (FRET) [181,182,183]. The optical properties of SWCNTs are highly dependent on their chirality, as the chiral index (n, m) determines the wavelength of the near-infrared fluorescent emissions. Therefore, different kinds of SWCNTs will have their own characteristic spectra, which can facilitate multiplex detection of biomarkers without any significant cross-talk [184]. The strong and broad absorption properties and sustained fluorescence in the NIR-II band linked to low biological autofluorescence further improve their potential for optical biosensing and increase detection sensitivity. The fluorescence emitted by SWCNTs remains highly stable during long measurement periods with negligible photobleaching and blinking. Besides their use in FRET-based systems, SWCNTs also effectively immobilize biomolecules and enhance the refractive index changes, making them suitable for SPR and SERS platforms [185,186,187].

Similarly, high-aspect-ratio nanowires and nanoneedles enhance the performance of optical biosensors through their large surface areas and strong light–matter interactions. Through waveguiding, light scattering, and plasmonic coupling, nanowires (such as ZnO, Si, and GaN) provide quick, label-free biomolecule identification. In the meantime, nanoneedles can gently penetrate cells to monitor the unperturbed intracellular environment. These structures are ideal for single-cell analysis, wearable biosensors, and ultrasensitive nanoscale diagnostics because they serve as subwavelength waveguides and plasmonic antennas, amplifying fluorescence, FRET, and SERS signals to map intracellular pH, enzymatic activity, and biomarkers [188,189,190]. These distinctive optical properties have enabled the use of 1-D nanomaterials for ultrasensitive optical detection of disease biomarkers, including those related to asthma. For example, SiO₂-coated gallium phosphide (GaP) nanowires have been studied for the detection of the asthma biomarker IL-5 due to their effective light-guiding properties. With their ability to guide fluorescence to their ends, these nanowires produce signals that are more intense and show better contrast and signal-to-noise ratios compared to those produced by the conventional flat substrates and maintain detectable signals even at low serum concentrations. Compared with plasmonic gold nanorod-based sensors, which rely on SERS hot-spot formation, GaP nanowires may offer better signal uniformity and reproducibility because their performance is less dependent on nanoparticle aggregation. However, nanowire-based assays usually demand fluorescent labelling and specialized fabrication, whereas plasmonic SERS platforms often produce enhanced signals and lower detection limits. These observations emphasize the balance between reproducibility and sensitivity in the design of one-dimensional optical biosensors [164,191].

6.1.2. Two-Dimensional Nanomaterials

Two-dimensional (2D) nanomaterials are ultrathin materials with a single nanoscale dimension and large lateral dimensions. An exemplary case is graphene, a one-atom-thick honeycomb lattice of sp²-bonded carbon. Graphene possesses remarkable optoelectronic properties, in addition to its extraordinary mechanical strength and thermal conductivity. It is an ideal choice for the non-destructive detection, since it has the potential to increase the sensitivity of optical biosensors through improved large surface area, high electron mobility, broadband and adjustable absorption [192,193]. Another advantage of graphene (including its derivatives, such as graphene oxide (GO) and reduced graphene oxide (rGO)) is associated with its excellent fluorescence quenching capability that can be used for label-free detection of biomarkers using the FRET technique [193]. In addition, graphene is a versatile foundation material for the construction of two-dimensional (2D) heterostructures, as its single-atom thickness and van der Waals interactions enable the stacking or integration of other 2D materials without lattice-matching constraints. By combining complementary layered materials, it is possible to overcome limitations such as the lack of a bandgap and to create multifunctional biosensors [194]. A representative example is a GO-based immunosensor for asthma biomarker IL-5, where quenching of GO’s intrinsic fluorescence was achieved by HRP-catalyzed DAB (3,3′-diaminobenzidine) polymerization. Anti-IL-5 antibodies immobilized on GO formed a sandwich complex with IL-5 and HRP-conjugates, with fluorescence quenching proportional to IL-5 concentration (~10 pg/mL LOD) in human serum, demonstrating high sensitivity and specificity. Though this platform has the advantage of strong signal amplification through enzymes, the use of the sandwich immunoassay principle and multiple enzymatic and incubation steps during the assays makes it difficult to achieve simplicity and point-of-care application [195]. Likewise, a GO-based aptasensor detected asthma-related IgE via FRET. In this system, a fluorescein isothiocyanate (FITC) labelled aptamer was initially quenched by GO, while fluorescence was restored after binding with IgE. In human serum, the sensor demonstrated high sensitivity with a detection limit of 22 pM and good selectivity. Compared with antibody-based systems, aptamer-based sensors are simpler and easier to modify without the need for enzyme-mediated signal amplification, but then again, the efficiency of aptamers may differ depending on the sequence design and experimental conditions, affecting reproducibility [196].

Another example is the family of transition metal carbides and nitrides known as MXenes, which have the general formula M_n+1X_nT_x (where M is a transition metal, X is C or N, and T_x represents surface terminations like –O, –OH, or –F). Because of their surface functional groups, these compounds have exceptional metallic conductivity, mechanical flexibility, and hydrophilicity, as well as the ability to transfer electrons rapidly [197]. These functional groups also promote dipole and interfacial polarization that contribute to their tunable electrical and dielectric properties and further improve signal transduction. Their layered structure and tunable thickness also provide abundant sites for surface functionalization and biomolecule conjugation. Among its unique optical properties are broad UV-to-near-infrared absorption, high optical transparency, intrinsic photoluminescence, fluorescence-quenching capability, and high refractive index sensitivity, all of which help in sensitive biomarker detection. Their charge-transfer-enabled electronic structure and abundant surface functional groups improve SPR and SERS-based sensing [198,199]. In addition, their strong light–matter interactions, intrinsic peroxidase-like nanozyme activity and fluorescence-quenching capability allow biomarker detection on SPR, SERS, colorimetric, electrochemiluminescence and FRET-based platforms. These characteristics, along with their strong physicochemical stability and biocompatibility, make MXenes highly attractive for advanced optical biosensing applications [198,200].

6.1.3. Nanoparticle-Based Systems

Gold nanoparticles (AuNPs) are widely used in biosensors because of their strong plasmonic properties, high electrochemical sensitivity, versatile surface functionalization, and excellent biocompatibility. Their ability to enhance optical signals has made them valuable components of SPR- and SERS-based biosensing platforms, enabling highly sensitive biomarker detection [188,201]. In particular, AuNPs serve as efficient substrates for SERS assays, where Raman-active substances like 2-naphthalenethiol (2-NT) and 4-acetamidothiophenol (4-AATP) are frequently utilized as reporter probes, utilizing strong thiol-gold adsorption and distinctive vibrational signatures to provide long-lasting, ultrasensitive, and multiplexed biosensing [202]. Apart from AuNPs, the potential use of silver nanoparticles (AgNPs) in biosensing has been recognized due to their electrical conductivity and rapid electron transfer rate, which improve electrochemical sensors’ performance [203]. Their strong plasmonic properties support colorimetric, SPR-, and SERS-based sensing, and offer better signal amplification since they have sharper LSPR bands and lower energy losses than AuNPs. Furthermore, their high molar absorptivity and fluorescence-quenching ability have facilitated their application in FRET-based and metal-enhanced fluorescence (MEF)-based sensing platforms [201,204].

Iron oxide nanoparticles (IONPs), especially Fe₃O₄, exhibit superparamagnetism and catalytic behaviour, providing more advantages for optical biosensors. The adjustable surface chemistry, along with high surface area, makes them ideal candidates for enzyme immobilization without compromising biomolecular stability and while also preventing the formation of insulating bilayers. In optical-based sensors, IONPs enhance detection sensitivity by increasing the refractive index effect in SPR and can also be coupled with gold nanoparticles or quantum dots for SERS and fluorescence applications. In addition, they help separate particles from sample matrices more quickly while also promoting faster electron transfer. These combined magnetic and optoelectronic characteristics contribute to making IONP-based sensors ultrasensitive in detecting target molecules at femtomolar concentrations with high signal-to-noise ratios [205,206]. For example, one study utilized FRET-based fluorescence biosensing platform using Fe/Fe₃O₄ core–shell nanoparticles with peptide-aptamer recognition sequences for detecting asthma-associated biomarkers, including IL-6 and CCL20, in exhaled breath condensates (EBC). This platform achieved femtomolar-level limits of detection. The measured concentration ranges clearly distinguished healthy individuals from asthma patients, with significantly increased cytokine levels observed in diseased samples. IL-6 concentration level was not detected or was lower than ~10⁻¹⁵ mol L⁻¹ in healthy samples, whereas asthma patients showed significantly high concentrations ranging from approximately 1.5 × 10⁻¹⁰ to 5.0 × 10⁻⁹ mol L⁻¹. Although the capability of detecting biomarkers without the need for invasive procedures constitutes one major benefit offered by EBC over other techniques, the effectiveness of the system still needs to be proven through additional studies using large populations [207].

In addition to these platforms, quantum dots (QDs) are a useful class of nanoscale semiconductor materials due to their specific properties, including size-dependent fluorescence, high quantum yields, broad absorption, and excellent photostability. Quantum dots exhibit quantum confinement, resulting in discrete energy levels and tunable band gaps that enable precise control over sensing performance. QDs are also suitable for point-of-care testing (POCT) and resource-constrained settings due to low detection limits, portability, and potential for miniaturization [208]. They can be used as sensitive probes in biosensors due to their capacity to absorb broad-spectrum light and emit narrow, size-dependent spectra, operating through mechanisms like FRET, bioluminescence resonance energy transfer (BRET), chemiluminescence resonance energy transfer (CRET), photo-induced electron transfer (PET), and internal filtration effects (IFE) [192,209]. Many studies have shown that QDs can be used to identify various biomarkers. For example, a CdSe@ZnS quantum dot-based fluorescent immunochromatographic test strip was created to detect the asthma-related cytokine IL-6 at the point of care. The strip allows for detection in less than 20 min and may be analysed using a variety of methods, including a portable immunoanalyzer, a commercial immunoanalyzer, smartphone photography with ImageJ 1.x, or a dedicated smartphone application. The assay demonstrated high sensitivity (LOD: 2.65 pg mL⁻¹) and good specificity toward IL-6, while smartphone-based analysis improved portability and accessibility for point-of-care testing. However, the smartphone application produced less consistent results and lower sensitivity than ImageJ and commercial analyzers suggesting that further optimization of image-processing algorithms and calibration process may help improve its performance [210]. Advanced SnS₂ QD–MXene platforms have also been shown to enhance signal amplification, allowing detection of salivary exosomes for asthma diagnosis [211]. The major nanomaterial classes, their optical sensing mechanisms, advantages, and reported applications for asthma biomarker detection are compiled in

Table 3. Nanomaterial-Enhanced Optical Biosensing Platforms for Asthma Biomarkers.

S. No.	Nanomaterial Class	Representative Materials	Primary Optical Mechanism	Advantages	Demonstrated Asthma Biomarker Application	LOD	References
1	1D nanomaterial	CNTs, Nanorods, Nanowires	Tunable LSPR, SERS Signal Amplification via shape-controlled hot spots, NIR photoluminescence	Photostable, allows multifunctional sensing, functionalizable, biocompatible, high sensitivity, label-free detection	IL-5	0.1–50 pg/mL	[164,177,182,212]
2	2D nanomaterials	Graphene, GO, rGO, MXenes	FRET quenching, SERS enhancement, and SPR refractive index change	excellent biocompatibility, versatility across multiple light ranges (UV–NIR), multiplexing possible, rapid and low-cost detection	IL-5	~10 pg/mL	[193,195,211,213]
					IgE	22 pM
					Exosomes (salivary)	2.5 × 10−14 g mL−1
3	Nanoparticle (NP) based system	Noble metal NPs-Gold (AuNPs), Silver (AgNPs)	LSPR, FRET quenching, ECL/CL catalysis, metal-enhanced fluorescence	Biocompatibility, easy functionalization, high Sensitivity, Simple color change for rapid testing without heavy equipment	IL-6	1.95 μg·mL−1	[214,215,216]
					IgE	10 pg/mL
		Magentic NPs- Iron Oxide (Fe3O4)	SPR, fluorescence with refractive index enhancement	High sensitivity, magnetic enrichment of analytes, reduced background interference	IL-6	~<1 × 10−15 mol L−1	[207,217]
		Quantum Dots	FRET, BRET, CRET	High fluorescence intensity, photostability, tunable emission and multiplexing	IL-6	2.65 to 50 pg/mL	[192,210,218]

.

6.2. AI Integration Biosensor

Artificial intelligence (AI) can be integrated with a biosensor for automated interpretation of complex multidimensional sensing data and enhancing sensitivity, specificity, and real-time disease detection. AI also contributes to better biosensor performance and clinical decision-making by optimizing data acquisition, noise filtering, pattern recognition, predictive diagnostics and continuous monitoring through wearable or implantable devices for early disease prediction and progression assessment. AI approaches that include machine learning and deep learning algorithms like support vector machines (SVM), random forest, K-nearest neighbor (KNN), naïve Bayes, convolutional neural networks (CNN), and recurrent neural networks (RNN) help in identifying patterns, classifying disease states, measuring biomarkers, and improving the accuracy of analysis [219,220,221]. SVMs use kernel functions to map the input biosensor features into a high-dimensional space, whereby the support vectors define the boundary that separates the two classes, and an optimal hyperplane is created to maximize the margin between the two [222]. Random Forest improves predictive performance using a combination of decision trees and making final recommendations by majority vote, which helps reduce overfitting and makes it more resistant to noise. K-nearest neighbors (k-NN) is a simple distance-based machine learning algorithm that classifies unknown biosensor samples by calculating distances between the samples in feature space and finding the closest training samples. The sample is then assigned to the class that appears most often in the k nearest neighbors [223]. Naïve Bayes uses probabilistic reasoning based on Bayes’ theorem for predicting the probability that a sample belongs to a particular class. It assumes the features of input to be conditionally independent of each other, which simplifies computation and allows efficient classification of biosensor data [224]. CNNs are capable of analyzing biosensor signals by using convolutional filters that scan small parts of the input in order to find useful patterns in the signal. Using multiple convolutional layers, CNNs learn hierarchical feature representations from simple signal patterns at shallow layers to more complex biomarker patterns at deeper layers. Pooling reduces the dimensionality of the signals but retains important features and increases noise tolerance, thus achieving efficient pattern recognition and classification. Whereas RNNs handle sequential biosensor data by remembering the hidden states containing the information from previous time steps. Due to their gated architecture, RNNs can control the flow of information in terms of what to retain, update, or ignore, hence making it possible to learn long-term temporal dependencies, which would help in modelling the time-varying changes in signal intensity and biological marker concentrations [225,226].

AI is used not only to interpret the data obtained as output, but also at various stages of biosensor design and development. It can assist in choosing appropriate analytes by identifying biomarkers using omics data and machine learning methods, and in designing recognition elements, such as antibodies and aptamers that have higher sensitivity and specificity, while also supporting material selection, transducer optimization, and sensor miniaturization [21,220]. At the material level, AI algorithms can establish structure-property-performance relationships and predict how biomolecules and sensing materials will behave under varying environmental conditions, including changes in pH and temperature [220]. As the performance of sensors relies on the complex interplay among materials used, nanostructure architecture, fabrication processes, environmental factors, and immobilization strategies, conventional trial-and-error optimization can be time-consuming and resource-intensive. To address this challenge, AI technology can simultaneously evaluate multiple design and fabrication variables, including nanomaterial concentration, incubation conditions, probe loading and signal-enhancement treatments to determine the best sensor design. Additionally, genetic algorithms have been used to optimize plasmonic nanostructures, improving sensitivity, detection range, and operational flexibility. Optimal conditions for biorecognition element immobilization such as pH, temperature, interfacial conditions and incubation time, can be selected with the help of AI to enhance the performance of biosensors by maximising binding efficiency, signal stability, and reproducibility [227,228].

In the context of optical biosensors, AI allows automated analysis of fluorescence, colorimetric, SPR, Raman, and photonic crystal signals. Machine learning algorithms correlate optical responses with analyte concentration and classify disease conditions. For example, in fluorescence-based biosensors, AI models analyze emission intensity, spectral shifts, and imaging data to allow sensitive and multiplex biomarker detection. Integration with smartphone platforms and cloud computing further enables real-time, sensitive, and reliable optical biosensing for disease biomarker detection [229]. Al-Khamees et al. proposed an AI-based asthma detection framework that combined respiratory sounds, wearable sensor data, environmental factors, and patient information using CNN, LSTM, and hybrid CNN–RNN models [230]. The system achieved 87.5% accuracy, with an F1-score of 86.5%, indicating that multimodal AI approaches can support monitoring and early detection. One of the major advantages of this study is that the researchers considered various physiological and environmental parameters that better represent the complex nature of asthma and support real-time monitoring through IoT-enabled wearable devices. With an AUC of 0.76 and both false-positive and false-negative predictions, the reported performance is still moderate, which may restrict its clinical reliability, particularly in missed asthma cases. Similarly, CNN-RNN-based pediatric monitoring systems indicated a diagnosis accuracy of roughly 94–95% for identifying irregular breathing patterns [231]. Breathomics-based nanosensor platforms also demonstrated positive results. For example, Ag-decorated ZnO chemiresistive nanosensors integrated with SVM-based pattern recognition enabled the detection of exhaled NO and H₂S for asthma diagnosis, achieving ppb-level sensitivity and approximately 81% classification accuracy in clinical breath samples [16]. By examining NO₂-related breath patterns, Cu₃(HHTP)₂-decorated Ti-MXene-based wearable nanosensors, combined with CNN, achieved about 97.6% recognition accuracy for asthma risk detection. The incorporation of CNNs facilitates automated identification of complex sensing patterns that may not be easy to detect using traditional analytical techniques, thereby improving classification performance. However, the current study has been conducted mainly in laboratory conditions, and the effectiveness of the proposed system in case of real environmental conditions, including various levels of humidity, temperature and breath contaminants, still requires further investigation [232]. Despite the advantages, several challenges remain significant for the clinical translation of AI-integrated biosensors, including low-quality or variable datasets, limited model adaptability, high computational demands, real-time processing constraints, and privacy and ethical concerns [219,233].

6.3. Smartphone-Based Biosensors

Smartphone-based biosensors combine biological sensing elements such as antibodies or enzymes with mobile technology to provide rapid, real-time, and low-cost diagnostics. In these devices, sensing may be done using optical or electrochemical methods, while the smartphone acts as a platform for signal processing, display, and wireless communication. To make them more sensitive and portable, approaches like wearable integration, nanomaterials, microfluidics, and AI-based mobile apps are being used that also support decentralized and point-of-care diagnostics [234]. In the case of smartphone-based optical biosensors, the built-in camera, light sources, and computing power of smartphones are used to record and analyze optical signals produced during sensing reactions. The detection can be done based on changes in absorbance, reflectance, fluorescence, luminescence, or surface plasmon resonance, which is then processed using smartphone applications to measure the quantity of the sample, often using color models such as RGB (Red, Green, Blue) or HSV (Hue, Saturation, Value). Additional components, such as LEDs, filters, or a dark box, can be attached to improve signal quality. Although fluorescence and chemiluminescence offer better sensitivity, colorimetric techniques are easy to perform, but are more dependent on ambient lighting conditions [235,236]. Sha et al. reported a smartphone-based optical sensor that detects hydrogen peroxide (H₂O₂) in exhaled breath, an indicator of airway inflammation in asthma [144]. This system uses a 3D-printed sensing device with controlled illumination and dye-based colorimetric reactions, where color shifts are captured by the smartphone camera, which then transforms them into RGB values for quantitative analysis. In order to estimate H₂O₂ concentrations, these signals are analysed via a mobile interface and integrated with machine-learning algorithms, allowing for quick and non-invasive breath monitoring. The performance is highly dependent on controlled lighting and camera conditions, which could make it harder to reproduce the same results outside a laboratory setup. At the same time, the machine-learning model is built on a relatively small dataset, so its behaviour across more diverse samples or patient groups is still uncertain. Multiple studies have also presented optical fibre–based wearable respiratory monitoring systems for continuous health assessment, underscoring their potential for affordable, flexible, and real-time health assessment. A smartphone-interfaced polymer optical fibre (POF) sensor system with IoT/cloud integration has been developed for respiratory rate monitoring, enabling real-time measurement and remote access via mobile-based signal processing. Experimental validation revealed respiration rates of 30.36, 35.88, and 41.4 BPM in comparison to reference metronome-controlled values, demonstrating the system’s great accuracy with errors of less than 2% [20]. Similarly, a textile-embedded multi-core fibre sensor has shown stable functionality irrespective of varied breathing styles and bodily states, with a Pearson correlation coefficient exceeding 0.9 against commercial systems [237]. Furthermore, a completely textile-integrated optical fibre sensor based on intensity modulation has been created to track respiration and differentiate between various breathing patterns based on the location of the sensor on the human body, supporting long-term wearable monitoring [238]. Even though all platforms use data collected and processed through smartphones, there is variability in how effectively they work based on the method of transduction used. Fibre- and textile-deformation systems tend to produce more reliable results because of the direct mechanical coupling with respiration, whereas intensity-based methods are more sensitive to attachment conditions and motion-induced variations. Smartphones are also being incorporated into asthma management systems, such as Asthma Attack Monitoring Devices (AAMD), enabling remote data exchange with healthcare providers, continuous monitoring, and real-time alarms. These examples show how smartphone-based biosensing can help with early detection and prompt treatment in asthma care [239].

7. Clinical Translation Barriers

The translation of biosensor technologies into clinical practice may face obstacles due to financial barriers and poor accessibility to healthcare services. Even low-cost biosensors will be unaffordable in resource-limited settings. Besides the price factor, economic viability also depends on disease occurrence, medical costs, and increased diagnostic accuracy [240]. The laser-based devices can cost more than EUR 100,000, which makes it impossible to use them in everyday practice or point-of-care facilities, while chemiluminescence analyzers can cost between USD 20,000 and USD 45,000. Although electrochemical sensors such as NIOX MINO are comparatively portable and cheap, they require frequent replacement, which raises maintenance costs [241]. Despite lower prices for wearable sensors, the total expenses also cover hardware and software integration within healthcare infrastructures and user training. Therefore, evidence of their cost-effectiveness in the long run remains limited, calling for further study [242,243]. The process of translating biosensors from the lab bench to clinical practice often takes time because of retrospective and prospective clinical studies (involving a large number of patients and lasting for years), multi-site trials, large-scale manufacturing, and regulatory approvals [238,244]. Depending on the market segment, devices must comply with different regulatory frameworks such as the U.S. Food and Drug Administration (FDA), the European Medical Device Regulation (MDR) and In Vitro Diagnostic Regulation (IVDR), India’s Medical Device Rules (MDR 2017), and the Therapeutic Goods Administration (TGA) in Australia. The process of obtaining approval involves presenting comprehensive proof of scientific validity, analytical and clinical performance, reproducibility, safety, clinical usefulness, and manufacturing quality. In addition, the differences in regulatory pathways between countries, as well as the need for risk-based classification, quality management systems, and post-market surveillance, might add considerably to the time and resources needed for regulatory approval [244,245,246]. Thus, despite years of intensive research and success in laboratory studies, many biosensors have not yet been accepted into clinical practice and commercial markets. Although some asthma-related biosensors, mostly FeNO monitoring devices such as NIOX VERO, NIOX MINO, NObreath, NOA 280i, and HypAir FeNO, have achieved commercial success and clinical adoption, the majority of emerging biosensor platforms remain at the research or validation stage [241]. Apart from these limitations, the lack of validation across various patient populations represents another substantial research gap that obstructs the practical utilization of biosensors in the medical setting. Although many new technologies have shown excellent sensitivity in controlled lab conditions, how well they work in different real-world environments has yet to be assessed. This lack of validation in different age groups and environments hampers the development of equitable, broadly applicable diagnostic tools [247]. One area where this problem occurs most often is paediatrics, where physiological and behavioural differences may affect biomarker detection. For example, pediatric populations, especially those in the preschool age range, often have difficulty maintaining constant breathing effort or volume, which makes it challenging to obtain uniform exhalation. Consequently, VOCs may get diluted or modified while collecting the sample, thus resulting in low sample quality and reduced reproducibility, limiting diagnostic reliability in clinical settings [248]. Similarly, the adoption of these biosensors is also determined by patient acceptance and product usability. They must be user-friendly, well-designed, and accessible to diverse users. The implementation of user-centred design, where patients and healthcare professionals participate in the development process, can improve correct usage, reliability, and confidence in the biosensor-based diagnostic tools [244]. The above aspects are especially important for wearable biosensors intended for long-term use in clinical and home-monitoring settings. While these devices are already in practice, prolonged use can cause skin irritation, and biofouling, environmental interference, and signal drift can affect their precision over time. The limited battery power and the lack of data security may also impede the device’s performance and its widespread acceptance because of poor user experience [249].

These barriers can be mitigated through a multidisciplinary approach that combines technical innovation with improvements to the healthcare system. It is important to note that the way forward to ensuring the success of biosensor devices in low-resource countries is through the adoption of low-cost, bottom-up design approaches using affordable materials and open-source components, along with public–private funding, value-based policies, streamlined regulations, expandable production, and trained healthcare personnel [240]. Additionally, new developments in semiconductor and microelectronics technology allow the making of compact, affordable, and low-power medical sensors. These technical leaps are increasingly supported by strategic partnerships, such as the Indo-German collaboration aimed at strengthening semiconductor manufacturing, which can support large-scale fabrication and commercialization of biosensors [250]. The successful incorporation of biosensor technologies into regular clinical practice will require strong data management systems, interoperability with existing healthcare infrastructures, and standardized protocols for data interpretation and reporting [251]. Ultimately, these limitations show the importance of clinical validation and technical optimization in order to connect laboratory research and clinical application in the field of asthma care.

8. Conclusions and Future Directions

The management of asthma is moving towards precise and rapid biomarker-based approaches due to the necessity to address the problem of disease heterogeneity and inflammatory endotypes, which cannot be identified using traditional diagnostic methods. While traditional diagnostics are still the foundation for diagnosing asthma, the lack of specificity and incapacity to capture the underlying biological complexity emphasize the need for more innovative alternatives. Current technological advancements in optical biosensors, supported by nanomaterial-based signal enhancement, offer promising routes for sensitive and non-invasive detection of asthma-related biomarkers in samples like breath and saliva. Alongside optical systems, electrochemical, chemiresistive, FET-based, and piezoelectric biosensors are also quite promising for rapid, low-level detection suitable for point-of-care applications. However, most current systems are only in the early stages of development and have to be validated before clinical translation. The main challenges to the success of these platforms include limited stability in real biological environments, biofouling, and the absence of large-scale, multi-centre validation across diverse asthma phenotypes, particularly non-type-2 asthma. Moreover, reproducibility problems associated with the synthesis of nanomaterials and the lack of robustness of the AI algorithms trained on small datasets further constrains its real-world applicability. Future work should prioritise systematic validation of biomarkers for non-type-2 asthma, the poorly characterized and often misdiagnosed form of asthma, improving sensor stability and reproducibility under real-world conditions. Standardisation of fabrication and testing protocols will be essential to improve reproducibility across different sensing platforms. Parallel to this, large-scale clinical trials with different populations are necessary for the successful transition from prototypes in the laboratory to clinical devices. The use of biosensors with AI technologies that are already proven to work clinically can help in continuous and personalized disease monitoring. The cost and scalability of the platform will also determine how this technology is used in low-income nations and low-resource environments. Therefore, coordinated progress in materials science, clinical validation, and data analysis would be necessary for the successful development and implementation of optical biosensors into reliable tools for routine asthma diagnosis and management.