Search Results (1,063)

Continuous, non-invasive cardiopulmonary monitoring is receiving increasing attention as population aging and chronic diseases rise. Acoustic sensing provides diagnostically relevant information with relatively simple hardware. Yet, physiological body sounds span heterogeneous and partially overlapping spectra and are highly susceptible to environmental noise and motion artifacts, which limit conventional stethoscopes and fixed-frequency sensors. Frequency-Tunable Acoustic Sensors (FTAS) offer a promising route toward frequency-selective amplification and adaptive interference suppression by matching their resonance to target signals, thereby potentially supporting multi-site monitoring and personalized diagnostics on a single platform. This review starts with an overview of physiological sound generation and the evolution of auscultation, then surveys mainstream medical acoustic transducers (piezoelectric, capacitive microelectromechanical systems (MEMS), piezoresistive and triboelectric) and their limitations in frequency selectivity. Resonance-tuning strategies are classified into three paradigms: electrical tuning, material-based tuning, and geometric reconfiguration, and their tuning ranges, response characteristics, and representative implementations are comparatively discussed. Finally, this review discusses the potential translational value of FTAS in physiological acoustic signal monitoring, particularly in cardiovascular and respiratory assessment, and emphasizes the remaining challenges, including the trade-off between sensitivity and selectivity, as well as long-term biocompatibility. At the same time, this review highlights their development prospects in customizable acoustic sensing platforms. Full article

Kolmogorov–Arnold Networks (KANs) have recently gained increasing attention as an alternative to conventional neural architectures, mainly because they replace fixed activation functions with learnable univariate mappings defined along network edges. This design not only increases modeling flexibility but also makes it easier to interpret how inputs are transformed within the network while maintaining parameter efficiency. KANs are particularly well suited for sensor-driven systems where transparency, robustness, and computational constraints are critical. This study provides a survey of KAN-based approaches for processing sensor data. A literature review conducted from 2024 to 2026 examined the deployment of KAN models in industrial and mechanical sensing, medical and biomedical sensing, and remote sensing and environmental monitoring, utilizing a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)-based methodology. We first revisit the theoretical foundations of KANs and their main architectural variants, including spline-based, polynomial-based, monotonic, and hybrid formulations, to structure the discussion. From a practical standpoint, we then examine how KAN modules are integrated into modern deep learning pipelines, such as convolutional, recurrent, transformer-based, graph-based, and physics-informed architectures. KAN-based models demonstrate comparable predictive performance as conventional machine learning models, while having fewer parameters and more interpretable representations. Several limitations persist, including computational overhead, sensitivity to noisy signals, and resource-constrained device deployment challenges. Real-world sensor systems encounter significant challenges in adopting KAN-based models, including scalability in large-scale sensor networks, integration with hardware architectures, automated model development, resilience to out-of-distribution conditions, and the need for standardized evaluation metrics. Collectively, these observations provide a clearer understanding of the current and potential limitations of KAN-based models, offering practical guidance on the development of interpretable and efficient learning systems for future sensor equipment applications. Full article

Multimodal biomedical machine learning increasingly integrates heterogeneous data sources (including medical imaging, multi-omics profiles, electronic health records, and wearable sensor signals) to support clinical diagnosis, prognosis, and treatment response prediction. Despite strong empirical performance, most existing multimodal systems lack a principled theoretical foundation for understanding why fusion improves prediction, how information is distributed across modalities, and when models can be trusted under incomplete or shifting data. This paper develops a unified information-theoretic framework that formalizes multimodal biomedical learning as an information optimization problem. We formulate multimodal representation learning through the information bottleneck principle, deriving a variational objective that balances predictive sufficiency against informational compression in an architecture-agnostic manner. Building on this foundation, we introduce information-theoretic tools for decomposing modality contributions via conditional mutual information, quantifying redundancy and synergy, and diagnosing fusion collapse. We further show that robustness to missing modalities can be cast as an information consistency problem and extend the framework to longitudinal disease modeling through transfer entropy and sequential information bottleneck objectives. Applications to multimodal foundation models, uncertainty quantification, calibration, and out-of-distribution detection are developed. Empirical case studies across three biomedical datasets (TCGA breast cancer multi-omics, TCGA glioma clinical-plus-molecular data, and OASIS-2 longitudinal Alzheimer’s data) show that the framework’s key quantities are computable and interpretable on real data: MI decomposition identifies modality dominance and redundancy; the VMIB traces a compression–prediction tradeoff in the information plane; entropy-based selective prediction raises accuracy from 0.787 to 0.939 at 50% coverage; transfer entropy reveals stage-dependent modality influence in disease progression; and pretraining/adaptation diagnostics distinguish efficient from wasteful fine-tuning strategies. Together, these results develop entropy and mutual information as organizing principles for the design, analysis, and evaluation of multimodal biomedical AI systems. Full article

Nonsteroidal anti-inflammatory drugs such as phenylbutazone (PBZ) are among the most widely used medications globally due to their effectiveness in relieving pain and reducing inflammation. This study aims to detect PBZ with metallic particle-based electrochemical sensors using cyclic voltammetry (CV) in the presence of catechol as a redox probe. The approach focuses on evaluating the electrochemical behaviour of PBZ under different experimental conditions and optimizing the detection parameters to develop a simple, rapid, and cost-effective analytical method suitable for this pharmaceutical compound in lab practice. CV was performed using four types of screen-printed electrodes, each modified with different transitional metal particles, in potassium ferrocyanide/potassium ferricyanide, catechol, and catechol-PBZ solutions to study the electrochemical response and detection capability for PBZ. The best performance characteristics were obtained for the sensor modified with Ir particles that detect PBZ, with a linearity range of 0.01 to 1.00 μM and a detection limit of 1.53 nM. Additionally, Fourier-transform infrared spectroscopy (FT-IR) was used to characterize the PBZ in pharmaceuticals. The method using an iridium-modified sensor developed in this study allows the accurate detection of PBZ in pharmaceuticals with a relative error lower than 4%. Full article

This narrative critical review addresses class imbalance in medical imaging, particularly within the context of multi-sensor and multi-modal environments, poses a critical challenge to developing reliable AI diagnostic systems. The integration of heterogeneous data from sources like CT, MRI, and PET presents a unique opportunity to address data scarcity for rare conditions through fusion techniques. This review provides a structured analysis of strategies to tackle class imbalance, categorizing them into data-centric (e.g., advanced resampling like SMOTE-ENC for mixed data types, GAN-based synthesis) and model-centric (e.g., loss function engineering, transfer learning, and ensemble methods) approaches. Crucially, we highlight how multi-sensor feature fusion and decision-level fusion paradigms can inherently enrich representations for minority classes, offering a powerful frontier beyond single-modality learning. We evaluate each method’s merits, clinical viability, and compliance considerations (e.g., FDA). Finally, we identify emerging trends where imbalance-aware learning synergizes with multi-sensor fusion frameworks, federated learning, and explainable AI, charting a roadmap toward robust, equitable, and clinically deployable diagnostic tools. Our quantitative synthesis shows that data-centric strategies can improve minority class recall by 12–35% in datasets with imbalance ratios (majority:minority) ≥10:1, while model-centric strategies achieve an average AUC improvement of 0.08–0.21 in multi-sensor medical imaging tasks with sample sizes ranging from 50 to 50,000. Full article

Artificial Intelligence (AI) has strong potential in health monitoring systems to support high-quality healthcare while mitigating cybersecurity risks. AI-based solutions for health and wellness applications, particularly for cardiovascular disease monitoring, are being explored to address complex healthcare challenges and improve patient outcomes. The integration of quantum and AI-based techniques is also gaining attention for enhancing future healthcare applications and communication technologies. Purpose: The primary objective is to improve cardiac care by accurately predicting symptoms and mitigating cyber-risks that threaten digital health integrity. By leveraging Integrated Quantum Networks (IQNs) and AI-driven protocols, this research aims to reduce the prevalence/incidence of non-communicable diseases by 50% by 2035 through proactive prevention and superior treatment management. Method: The framework utilizes AI-based techniques and AI-quantum-enhanced sensors and IQN to build a secure, proactive monitoring system. This theoretical framework integrates high-precision data collection with robust risk management systems to protect against vulnerabilities in digital health infrastructure. These components work in tandem to ensure that sensitive medical data remain resilient against emerging cyber threats. Anticipated Results and Conclusions: The system is expected to improve cybersecurity resilience, system performance, and energy efficiency (EE), supporting the development of secure and advanced future healthcare applications. Full article

Although natural enzymes have a high catalytic activity as biocatalysts, they still face many limitations in practical applications, including high preparation and purification costs, poor environmental stability, and difficulties in recovery and reuse. Nanozymes are a class of synthetic nanomaterials with enzymatic catalytic properties. They are regarded as promising alternatives to natural enzymes due to their low cost, good stability, adjustable catalytic activity, and easy surface modification. Among many nanozyme materials, metal–organic frameworks (MOFs) and covalent organic frameworks (COFs) have attracted much attention due to their high specific surface area, adjustable porosity, and stable framework structure. This review summarizes the latest research progress of nanozymes based on MOFs and COFs and reveals the catalytic properties of different enzymes (oxidase, peroxidase, catalase, glucose oxidase, superoxide dismutase, hydrolase) simulated by them. In addition, their potential applications in sensors and medical fields are discussed. Finally, this review discusses the current challenges and developments of organic framework-based nanozymes and provides suggestions for future research directions. Full article

Biocompatible thin films are essential for advancing biomedical devices, as they enhance integration with biological tissues, improve device longevity, and reduce complications. The rapid evolution of both medical needs and materials science has led to a diverse array of deposition techniques, each offering unique advantages and challenges for tailoring surface properties without compromising the bulk characteristics of implants and sensors. While laser-based methods—such as pulsed laser deposition (PLD) and Matrix-Assisted Pulsed Laser Evaporation (MAPLE)—are renowned for their precision, ability to preserve complex material stoichiometry, and suitability for low-temperature processing, the broader landscape includes several other important approaches. Physical Vapor Deposition (PVD) techniques, including magnetron sputtering and pulsed electron deposition, are widely used for their ability to create uniform, adherent coatings with controlled thickness and composition, making them suitable for both hard and soft biomedical substrates. Chemical Vapor Deposition (CVD) and its plasma-enhanced variant (PECVD) offer conformal coatings and excellent control over film chemistry, which is particularly valuable for functional polymer and ceramic films. Other methods, such as sol–gel processing, ion beam deposition, and electrophoretic deposition, provide additional flexibility in terms of coating composition, adhesion, and processing temperature, allowing for the fabrication of films with tailored mechanical, chemical, and biological properties. Despite these advances, the field faces ongoing challenges in optimizing film properties for specific clinical applications, ensuring reproducibility, and scaling up production for widespread use. The necessity of this review lies in its comprehensive comparison of laser-based techniques with alternative deposition methods, providing critical insights into their respective strengths, limitations, and suitability for different biomedical scenarios. By synthesizing recent developments and highlighting current gaps, this review aims to guide researchers and clinicians in selecting the most appropriate thin-film deposition strategies to meet the evolving demands of next-generation biomedical devices. Full article

Background/Objectives: Effective glycemic management from the time of diagnosis is essential in the care of children and adolescents with type 1 diabetes (T1D), as early glycemic patterns can influence long-term health outcomes. Methods: This exploratory study evaluated a one-month interactive, group- and play-based education program designed to enhance food and carbohydrate counting skills among families of children and adolescents with newly diagnosed (ND) T1D (<1 year since diagnosis) or suboptimal glycemic control (SGC) (hemoglobin A1c (HbA1c) > 7.5% (58 mmol/mol)). The intervention included hands-on learning activities in food and carbohydrate counting, and peer interaction to support development of diabetes self-management skills. Data were collected at baseline, post-intervention, and at six-months follow-up through medical records, glucose sensor data, and a questionnaire assessing diabetes self-management skills, dietary practices, and carbohydrate counting. Results: Between September 2022 and April 2024, 55 children and adolescents were enrolled in the ND group and 22 in the SGC group. Post-intervention, carbohydrate counting skills improved, particularly in the ND group. Participants reported greater confidence and independence in carbohydrate counting and insulin dosing, with parents noting sustained benefits at six-months follow-up. No significant changes were observed in glycemic control, including time-in-range and postprandial glucose profiles. Conclusions: In this exploratory study, early interactive and play-based group education was associated with improvements in carbohydrate counting skills and self-care confidence in children and adolescents with newly diagnosed T1D. These improvements were not accompanied by changes in glycemic outcomes. The findings occurred during a complex and transitional phase following diagnosis. Further research is needed to examine sustainability and long-term clinical impact. Full article

With the advent of ubiquitous healthcare and advancements in textile industry, non-invasive wearable biomedical solutions are becoming an increasingly attractive alternative to in-hospital monitoring, allowing for timely diagnostics and prediction of severe medical conditions. Non-contact biopotential monitoring is particularly promising because non-contact biopotential electrodes can be applied over clothing or embedded in the material without almost any preparation. However, due to the intricacies of capacitive coupling they rely on, the design of such electrodes and their interface with the body plays a key role in achieving measurement repeatability and their widespread utilization in clinical-grade diagnostics. Based on exhaustive investigation of several decades of the literature on non-contact and capacitive biopotential electrodes and electric potential sensors, this study is intended to serve as a state-of-the-art overview of their historical development and design challenges, a collecting point for important research theories and development milestones, a starting point for anyone seeking for a soft head start into this research area, and a remedy for occasional misnomers and conceptual errors identified in the existing papers. The ultimate goal of this comprehensive analysis is to demystify phenomena of non-contact biopotential monitoring and capacitive coupling, systematically reconciliate terminological inconsistencies, and enhance accessibility to the most important findings for future research. To accomplish this, fundamental concepts are thoroughly revisited—from fundamentals of electrochemistry and working principles of capacitors and operational amplifiers to system stability and frequency-domain analysis. With the use of various mathematical tools (Laplace transform, phasors and Fourier analysis, and time-domain differential calculus), discussions on non-contact and capacitive biopotential electrodes, collected from the 1960s onward, are for the first time compiled into a unified, abstracted, bottom-up analysis. The laid-out inspection provides analytical explanation for various aspects of measurement results available in the referenced literature, but also serves an educative purpose by devising a methodological framework that can be easily applied to other similar research fields. Firstly, the differences and similarities between wet, dry, surface-contact, non-contact, capacitive, insulated, on-body, and off-body biopotential electrodes are clarified. For this purpose, equivalent electrical models of various non-invasive biopotential electrodes are analyzed and compared. As a result, a proposal for a revised classification of biopotential electrodes is given. Secondly, instead of using the concept of a purely capacitive biopotential electrode, a test is proposed for assessing the predominant coupling mechanism achieved with an electrode over an insulating layer. Thirdly, a fundamental model of a buffer active non-contact biopotential electrode and its interface with the body is built and generalized, and the proposed test is applied for analyzing the influence of voltage attenuation and phase shifts on signal morphology. Lastly, guidelines for designing the described electrode–body interfaces are proposed, along with a discussion on practical aspects of their implementation. Full article

The application of MXene–polymer composites to wearable and implantable medical devices requires the development of hydrophilic and biocompatible MXene–polymer hydrogel composites with high electromechanical response, flexibility, and durability. Here, we formulate low weight percentage MXene–hydrogel copolymer inks enabling the direct light processing (DLP) of Ti₃C₂T_x MXene–polyvinyl alcohol (PVA)–polyacrylic acid (PAA)–hydrogel composites. The low wt% MXene–PVA–PAA composites demonstrate high biocompatibility, mechanical flexibility, high sensitivity and high precision for sensing acute bending angles. The sub-millidegree angle resolution of these electromechanical sensors demonstrates their suitability for applications such as the highly precise tracking of joint movements. In addition, the synthesized MXene membranes show promise for applications in osmotic energy conversion, with a harvested electric power density of 6.79 Wm⁻². Full article

Accurate monitoring of intracranial pressure (ICP) is critical for the diagnosis and management of neurological disorders. Although various ICP sensors have been developed, their sensitivity is often limited, restricting their ability to detect subtle pressure variations. Therefore, there is a pressing need to develop ICP sensors with enhanced sensitivity to improve measurement accuracy and patient outcomes. In this paper, a highly sensitive and precise pressure sensor for intracranial pressure (ICP) monitoring was proposed. Theoretically, the beam-membrane-island structure was introduced and optimized to improve sensitivity and linearity compared to a flat membrane structure. The notches etched at beam end were designed for further improving sensitivity. Experimentally, the designed sensor achieved a sensitivity of 1.59 mV/V//kPa and a nonlinearity of −0.22% F.S. Additionally, the sensor can detect pressure with centimeter water column (cm H₂O) resolution, making it suitable for ICP monitoring. This technology holds broad application prospects in the field of medical devices. Full article

Severe Dengue fever can cause Dengue Hemorrhagic Fever (DHF), a life-threatening condition characterized by plasma leakage and hemoconcentration. A hematocrit (Hct) rise of ≥20% is a key indicator for medical intervention, but current monitoring is invasive and intermittent. This study aims to determine the optimal design parameters for a non-invasive optical sensor to continuously monitor hemoconcentration. We developed a high-fidelity Monte Carlo model of light transport in a multi-layered skin model, with the epidermis set to a 5% melanin volume fraction (Fitzpatrick type II/III). To ensure signal reliability, simulations were conducted with a high photon count ($1\times {10}^8$ photons), yielding a stochastic (Monte Carlo) signal-to-noise ratio of approximately 36 dB. We simulated diffuse reflectance at four characteristic wavelengths (577 nm, 660 nm, 800 nm—the isosbestic point—, and 940 nm) over source-detector separations of 0.5–8.0 mm. Sensor sensitivity was quantified as the reflectance change for a +25% relative Hct rise (e.g., 42% to 52.5%), mimicking severe hemoconcentration, and its dependence on baseline dermal blood volume fraction (BVF) was investigated. Sensor sensitivity showed a non-linear dependence on BVF, showing a direct correlation with perfusion level, reaching an optimal 6.41% for a robust 5% BVF at 8.0 mm. A dedicated sweep showed that even under low-perfusion shock conditions (1% BVF), the sensor maintains a highly significant sensitivity of 5.71% (also at 8.0 mm), indicating that sensitivity remains high across a physiologically relevant perfusion range. In the analysis, at a robust 5% BVF, the 800 nm wavelength demonstrated superior reliability, with peak sensitivity at 6.41% at 8.0 mm. Visible wavelengths (577 nm and 660 nm) exhibited high theoretical sensitivity, while 940 nm was compromised by water absorption. Based on these findings, a non-invasive optical sensor for hemoconcentration is most effective operating at 800 nm, within the evaluated spectral set, with a source-detector separation of ≥6.0 mm, targeting the deep dermis while minimizing superficial interference. This design provides an optimal balance of tissue penetration, robust sensitivity to Hct changes, and reduced sensitivity to oxygenation-related variability while maintaining signal stability. This work enables the design of a device for continuous monitoring, supporting continuous monitoring of hemoconcentration trends relevant to plasma leakage progression. Full article

The high rate of bike commuting around the globe has greatly transformed the mode of transportation in cities, but the high speeds of motorized cycling have contributed to a high rate of serious road trauma. Although conventional helmets offer necessary passive structural protection, they do not consider the most important aspect of the emergency response, which is the Golden Hour the time frame during which medical intervention can have the most significant impact. This paper is a development and validation of an autonomous, low-cost smart helmet architecture that is programmed to operate in real-time to detect accidents and autonomously inform the operator of accidents. The system is built up of an ESP32 microcontroller with a multi-modal sensor package, which comprises an inertial measurement unit (IMU), force-impact sensors, and MQ-3 alcohol sensors to conduct proactive safety screening. To overcome the single threshold limitation of unreliable systems, a time-windowed sensor-fusion algorithm was applied in order to distinguish between normal riding dynamics and bona fide collisions. This reasoning involves concurrent cues of high-G inertial rotations and physical impacting features over a time window of 500 ms to reduce spurious activations. The architecture of the system is completely self-sufficient and employs an in-built GPS-GSM module to send the geographical location through SMS without the need to have a smartphone connection. The prototype was also put through 150 experimental tests, with some conducted in laboratories, and real-world running tests in diverse terrains. The findings reveal an accuracy in detection of 93.7, a false positive rate (FPR) of 2.6 and a mean emergency alert latency of 2.8 s. In addition, it was found that structural integrity was confirmed at ECE 22.05 impact conditions using Finite Element Analysis (FEA), with a safety factor of 1.38. These quantitative results mean that the proposed system is an effective way to address a cultural shift between passive structural protection and active rescue intervention as a statistical and computationally efficient safety measure of modern micro-mobility. Full article

Electrochemical biosensors are becoming increasingly prevalent across medical, food, and bioprocessing industries for monitoring complex biological processes. However, their sensitivity to contamination and exposure to potentially hazardous biological species often necessitates single-use disposal, contributing to the release of high-value, high-demand, and environmentally damaging materials into the environment. This study investigates the feasibility of a closed-loop recycling process for single-use glucose biosensors, with a focus on the recovery and reuse of noble metals silver and gold. Guided by ecodesign principles and using low-impact materials, we developed a silver screen ink, gold syringe ink, and a poly(lactic acid) (PLA) substrate. Sensors were fabricated by additive manufacturing and screen printing—enabling the scalability afforded by screen printing to produce the high-coverage silver layer while also minimising gold ink waste using additive manufacturing. A low-energy recovery method that exploited selective solvent compatibility was developed to reclaim silver and gold. Second-generation devices were then fabricated, demonstrating performance comparable to commercial equivalents while achieving an 80% reduction in material usage, cost, and environmental impact across 16 categories using a life cycle assessment (LCA). Full article