Cognitive burnout and chronic stress pose significant threats to long-term health and productivity, creating an urgent need for continuous, non-invasive physiological monitoring. This paper presents the design and implementation of an intelligent, edge-computed cyber-physical wearable system engineered for real-time stress tracking and burnout prediction. The hardware architecture leverages an ESP32-S3 microcontroller to interface with a physical sensing layer comprising a MAX30102 pulse oximeter and a Galvanic Skin Response (GSR) sensor. Raw photoplethysmography (PPG) and skin conductance metrics are sampled, conditioned, and subjected to localized feature extraction on the device. To eliminate latency, bandwidth constraints, and privacy vulnerabilities associated with continuous cloud reliance, a lightweight hybrid machine learning model consisting of Long Short-Term Memory (LSTM) networks and Graph Convolutional Networks (GCN) is deployed directly on the microcontroller using TensorFlow Lite Micro.

The LSTM layers isolate temporal, time-series trends in the biometric data, while the GCN captures the complex, interconnected Rachitha A.B Department of Electronics and Communication Engineering KS School of Engineering and Management Bengaluru, India E-mail: abrachitha4@gmail.com progressively rather than overnight, early detection is critical. Traditional methods of assessment heavily rely on periodic, subjective self-report questionnaires (such as the Maslach Burnout Inventory), which are inherently retrospective, prone to user bias, and incapable of providing timely, preventative interventions. To address these limitations, contemporary research has pivoted toward continuous, objective physiological monitoring.The human body’s autonomous nervous system reacts dynamically to cognitive strain, leaving distinct biometric signatures. Two of the most reliable and non invasive indicators physiological dependencies between cardiovascular and electrodermal responses. Experimental framework evaluations indicate that the system successfully identifies cognitive strain thresholds on-device, dispatching instantaneous "Burnout Alerts" to a user-facing mobile application dashboard while asynchronously syncing historical telemetry to the cloud for longitudinal health analytics. The proposed solution offers a highly scalable, secure, and energy efficient paradigm for pervasive health monitoring and early burnout intervention. of psychological stress are cardiovascular dynamics—tracked via heart rate variability (HRV) derived from photoplethysmography (PPG)—and electrodermal activity (EDA), measured through skin conductance. Capturing these signals simultaneously offers a multi-modal window into a user’s physiological state. Historically, processing such complex biomedical data required transmitting raw telemetry to resource-heavy cloud servers, introducing severe bottlenecks regarding data privacy, latency, and continuous power consumption.