Morning Anxiety Detection Through Smartphone-Based Photoplethysmography Signals Analysis Using Machine Learning Methods

Despite the absence of a standardized clinical definition, morning stress is widely recognized as the stress experienced upon waking. Given its established link to various diseases prevalent in modern society, the accurate measurement and effective management of stress are paramount for maintaining optimal health. In this study, we present a novel approach leveraging the sophisticated capabilities of smartphones to extract photoplethysmography (PPG) signals for immediate detection of morning stress. Data from 61 participants were meticulously collected and processed to extract PPG signals, subsequently employing 11 carefully selected features for stress detection. Through the utilization of the Support Vector Machine (SVM) for classification, we scrutinized the accuracy of our method against established benchmarks. Notably, by integrating the False Discovery Rate (FDR) formula and employing the Particle Swarm Optimization (PSO) algorithm, we achieved a significant enhancement in the classification rate, elevating it from 96% to an impressive 100%. These compelling results underscore the efficacy of our proposed methodology and illuminate the promising potential of smartphone-based morning stress detection as a viable tool for proactive health management.