REAL-TIME ADAPTIVE STRESS DETECTION USING PHYSIOLOGICAL SIGNALS FOR HIGH-RISK OPERATIONS

Accurate and real-time stress detection is essential to ensure safety and maintain peak performance in hazardous occupations. This study presents a tailored, real-time tension detection system that applies machine learning techniques to physiological inputs. The system collects data on heart rate variability (HRV), electrodermal activity (EDA), and skin temperature using wearable sensors. A machine learning procedure that involves feature extraction, normalization, and model training is used to dynamically classify stress levels. Personalized models are created for each individual using supervised learning techniques like support vector machines (SVM), Random Forests (RF), and sophisticated machine learning methods (CNN LSTM) to account for physiological differences. The system steadily increases accuracy by adjusting and learning from new data. Experiments carried out in hazardous simulation environments have demonstrated that the proposed method can identify stress in real time, offering crucial information for preventative measures. This study builds intelligent safety solutions for high-risk occupations using physiological sensing and adaptive machine learning.