Abstract
This paper presents the development and experimental validation of a wearable system for stress and cardiovascular monitoring that integrates three sensors: a Photo-Plethysmography (PPG) sensor, a Galvanic Skin Response (GSR) sensor, and a DS18B20 digital temperature sensor. The ESP32 microcontroller serves as the core of the system, performing signal filtering using the Exponential Moving Average (EMA) method and lightweight on-device classification with machine learning models (CNN, LSTM). Wireless communication is enabled using Wi-Fi and Bluetooth Low-Energy (BLE), allowing remote monitoring and cloud-based analytics. The system was tested on 10 volunteers under various physical and emotional scenarios, including sitting, walking, exercising, yoga, cycling, and watching movies. The results demonstrated significant physiological variations: the heart rate increased from 75 bpm at rest to 125 bpm during stress episodes, the skin conductance increased by up to 25%, and body temperature changed by 2-3°C depending on activity level. The embedded ML models achieved a stress classification accuracy of 92% (F1 = 0.90). The prototype also showed high energy efficiency, operating for more than 12 hours on a single charge. The proposed architecture offers promising applications for personalized health monitoring, stress management, and integration with medical information systems (FHIR, HL7).