Scalable Skin-Like Wearable Bioelectronics for Real-Time Physio-Cognitive Stress Monitoring in Ambulatory Settings

Introduction: In high-stress environments, monitoring and assessing physiological and cognitive stress is essential for enhancing performance in tasks requiring high concentration and preventing adverse health outcomes. Chronic stress, linked to significant health issues such as cardiovascular diseases and mental health disorders, affects over 450 million people globally and contributes to substantial healthcare costs. Here, we introduce a scalable, skin-like wearable platform that integrates advanced sensor technologies and machine learning algorithms for real-time, non-invasive stress monitoring in ambulatory settings. This dual-use on-skin platform is designed for continuous health monitoring both in clinical and everyday environments, fulfilling the need for effective stress management tools. 20 channels of bio-signal-based multi-modal machine learning framework have been developed for an automated clinical decision support (CDS) system, achieving 93% accuracy in binary stress and rest state classification and 84% accuracy in classifying various stress states across three distinct physiological protocols for 20 individuals. These advancements provide a continuous, non-invasive monitoring solution that promises 1) easy-to-use 20 channels driven real-time stress monitoring, 2) various physio-cognitive states monitoring capabilities, and 3) neural signal-driven medical diagnostics in ambulatory environments.

Materials and

Methods: The wearable platform comprises two main components: a cardiorespiratory (CR) patch worn on the chest and a neurological (Neuro) patch applied to the forehead. The CR patch incorporates sensors for electrocardiography (ECG) and seismocardiography (SCG) to monitor heart activities, enhanced by nanomembrane electrodes with layers of copper, chromium, and gold on a polyimide film, for better signal-to-noise ratio (SNR). Additionally, the design features serpentine interconnects engineered to reduce local strain and maintain mechanical stability, as demonstrated in computational modeling results. The chest-worn patch is also equipped with an accelerometer for capturing cardio-mechanical signals. The Neuro patch includes sensors for electroencephalography (EEG), electrooculography (EOG), electromyography (EMG), and electrodermal activity (EDA), utilizing FFC connectors for easy replacement of electrodes and backed by brown fabric tape with silicone gel for enhanced skin conformality. This ensures close skin contact to minimize data loss from motion and allows for high-quality signal transmission. Both patches are designed with flexibility and biocompatibility in mind, shown in tests demonstrating the patches' ability to withstand bending and stretching during continuous wear. Data collected from these patches are wirelessly transmitted to mobile and cloud systems and analyzed using a multimodal machine learning framework. This framework includes Convolutional Neural Networks (CNN) and multi-class Support Vector Machines (SVM), achieving an accuracy of over 93% in binary stress/rest state classification and 84% in classifying various stress states—mental fatigue, emotional stress, and physiological stress.

Results, Conclusions, and Discussions: This study presents a user-friendly, plug-and-play approach to stress monitoring using wearable bioelectronics, set to transform stress management practices across real-world scenarios. The development and validation of dual on-skin patches have successfully demonstrated the ability to continuously and non-invasively capture high-fidelity physiological and neurological signals under varying stress conditions. This integrated platform, which combines wearable technology with mobile and cloud computing, offers precise real-time assessment of multi-state stress, marking a significant advancement from traditional human-dependent methods to an automated clinical decision support system. By enabling early detection and timely intervention for stress-related health issues, this technology enhances our understanding and management of stress's impact on the body and further advances human assistive devices.

Acknowledgements (Optional): The authors acknowledge the support of the NBMC grant NB18-21-36, Center for Wearable Intelligent Systems and Healthcare (WISH), and IEN Center Grant from the Georgia Tech Institute for Electronics and Nanotechnology.