Predicting Stress Levels Through Physiological Signals with End-to-End Deep Learning Regression Models

Abstract

Assessing stress levels has traditionally relied on qualitative methods such as self-reports and clinical observations. With technological advancements, physiological indicators like heart rate, skin conductance, and cortisol levels have been used for more objective stress measurement. Machine learning has further improved stress assessment by enabling real-time analysis of complex physiological data.Thisstudyaimstodevelopanend-to-enddeeplearningregression model for accurate stress level prediction based on physiological signals. Conventional stress assessment methods, such as questionnairesandclinicalevaluations,oftenlackreal-timemonitoring capabilities, leading to subjective and delayed results. Traditional systems are also limited in their ability to analyze complex physiological data, affecting the effectiveness of stress management strategies. The proposed model will integrate multiple physiological signals, including heart rate variability, skin conductance, and respiratory rate, to enhance stress prediction accuracy. By leveraging deep learning techniques, the system will process real-time physiological inputs, enabling personalized stress management strategiesandinterventions.Thisapproachhasthepotentialtoimprove mental health monitoring, offering a more reliable and automated method for assessing stress levels.