TL;DR
The feasibility of integrating Edge AI with microfluidic biosensing concepts for low-latency and energy-aware monitoring of critical biomarkers is demonstrated and full analytical and device-level validation remains future work.
This work connects biomedical signal capture, on-device AI, clinical translation, and digital ethics to delineate a roadmap for frontier AI wearables, and proposed edge-aware transformer-based models, privacy-preserving analytics, and global multimodal sensing advance intelligent, scalable, and fair AI-driven health monitoring systems.
Wearable healthcare electronics have become widespread and rapid advancement of this technology has pushed for intelligent, real-time and non-invasive biosensing technologies. The traditional wearable sensors are generally low in sensitivity, susceptible to electromagnetic interference, and consume too much energy, which makes them unable to perform long-term healthcare monitoring. In this work, an Edge-AI enabled MXene-based fiber optic IoT biosensor architecture is proposed for intelligent wearable health care application. The proposed system combines MXene nanomaterial-coated fiber optic sensors, IoT communication, and edge artificial intelligence to facilitate real-time monitoring of biomedical signals and intelligent prediction of healthcare. Real-time analysis is performed on the biomedical signals acquired from the fiber optic biosensors by performing signal conditioning and feature extraction followed by lightweight classification algorithm on the edge-AI. This framework can enable low-latency communication, increased sensing sensitivity and energy efficient healthcare monitoring. Experimental results show the system’s improved signal stability, quick response speed and accurate health-state prediction when compared with traditional wearable sensing systems. The proposed framework has great potential in smart healthcare, Telemedicine, Wearable Electronics, and AI based remote patient monitoring applications.
The integration of smart nanosensor systems with artificial intelligence (AI) and the Internet of Things (IoT) is transforming diagnostic monitoring in biomedical, environmental, and industrial applications. These advancements facilitate real-time data acquisition, continuous physiological tracking, and point-of-care diagnostics, significantly enhancing early disease detection and personalized care. Innovations in nanomaterials, such as graphene, quantum dots, and plasmonic nanoparticles, have bolstered sensing capabilities, allowing for high sensitivity, rapid responses, and the selective detection of important biomarkers. However, challenges remain in translating these nanosensor technologies from the lab to real-world applications, including reliability, stability, integration with digital platforms, data security, power supply for ongoing monitoring, and clinical biocompatibility. Frequently, current systems operate in silos, lacking coordination between sensing, data processing, and decision-making components. This study proposes a systems integration framework that unifies smart nanosensors, AI-driven analytics, and the IoT into a single platform for continuous health monitoring. The main findings reveal that nanosensor-enabled systems can significantly enhance diagnostic performance, with graphene-based nanosensors detecting C-reactive protein in human serum at concentrations near 27 pM, and plasmonic sensors offering swift, portable detection. Additionally, electrochemical immunosensors achieve sensitivity for oral cancer biomarkers at concentrations as low as 1.424 fg mL-1 within just 120 seconds. The novel contribution of this review is the proposed systems-integration approach that unifies smart nanosensors, AI-driven analytics, and IoT into a cohesive platform for continuous health monitoring and data-driven decision-making in healthcare, showcasing the potential for improved diagnostic efficiency and patient outcomes. By proposing a multi-layered systems integration framework, this review synthesizes current advances in smart nanosensors, AI, and IoT into a unified architecture that addresses translational challenges and supports the development of intelligent, real-time, and patient-centered healthcare systems.
Clinical translation remains limited by data security and privacy risks, insufficient standardization and regulatory alignment, long-term stability and biocompatibility concerns, and uneven validation maturity across technologies.
By linking molecular-level recognition with computational signal interpretation, this review highlights the advantages and limitations of artificial intelligence-integrated biosensors for next-generation point-of-care diagnostics, continuous health monitoring, and personalized healthcare applications.
Smart wearable biosensors represent a significant paradigm shift from one-time sample analysis to real-time biochemical monitoring at the body interface. Besides the flexible design of the device or wireless readout, their clinical utility will also require the reliability of the entire sensing pathway under real physiological conditions. This pathway involves biofluid access to clinical interpretation. Despite rapid progress, many wearable biosensor platforms remain limited by weak biofluid–blood correlation, receptor degradation, biofouling, motion artefacts, sensor drift and insufficient patient-level validation. Thus, a chemistry-to-clinics approach is crucial to assess the analytical reliability and translational readiness of recognition elements, sensing materials, and engineered biointerfaces. Enzymes, antibodies, aptamers, nucleic-acid systems, molecularly imprinted polymers, and nanozymes are discussed within the context of selectivity, stability, antifouling behaviour and suitability for continuous monitoring of sweat, interstitial fluid, tears, wound exudate and breath condensate. The functionality of carbon nanostructures, metal-based nanomaterials, hydrogels, MXenes, metal–organic frameworks and self-powered interfaces are evaluated in terms of their applications in amplification, mechanical conformity, biofluid handling and signal stability. Artificial intelligence is positioned as a support layer for signal correction, calibration, classification, multimodal fusion and predictive interpretation, rather than as a substitute for robust sensing chemistry. This review provides a critical chemistry-to-clinical perspective on smart wearable biosensors and outlines the validation, manufacturing, cybersecurity, post-market surveillance and benchmarking requirements needed for their translation into reliable diagnostic and therapeutic-monitoring technologies.
MIT affiliates engage with the MIT-IBM Computing Research Lab to bring rigorous theory to production systems.
What if pathology foundation models could do more with less? GigaPath-Flash and GigaTIME-Flash cut computational demands while maintaining strong performance, opening the door to larger studies and broader exploration. The post GigaPath-Flash and GigaTIME-Flash: Toward population-scale discovery with efficient pathology foundation models appeared first on Microsoft Research.
With millions of users across the world, Julia has been used to conduct cutting-edge research and to design new drugs, jet engines, heat pumps, and more.
A new machine-learning framework aims to improve the success rate of computational protein design while moving away from results that reproduce sequences found in nature.
