Development of Deep Learning Based Stress Analysis Technique and Its Implementation Using Quality PPG

This article reports a processor implementation protocol of a personalize stress assessment technique. Owing to the increased prevalence of stress in modern society associated with its increased negative impact on human well-being, early intervention becomes critical. In this regard, wearable sensor enabled Biosignal based, especially Photoplethysmogram (PPG) based stress analysis has gained popularity among researchers due to its easy, cost-effective and non-invasive acquisition protocol. However, PPG is highly susceptible to motion artifacts and other environmental noises. Thus, signal quality analysis (SQA) becomes essential for any PPG based analysis. Very few studies have performed scalogram based stress analysis using PPG after assessing the quality of the biosignal. This study presents a signal quality analysed PPG signal based mental stress analysis using 2D scalograms. The study used PPG from publicly available WESAD dataset. Preprocessed PPG signals were passed through SQA block and beats obtained thereafter were transformed into scalograms using continuous wavelet transform (CWT). Scalograms were then fed into a convolutional neural network (CNN) for classification purpose. The results were further compared with those obtained without SQA and it was found that SQA enhanced the results.The entire algorithm was implemented on Nvidia Jetson Orin board and an accuracy of 93.98% was obtained using scalograms after SQA.