Towards enhancing the performance of a stress detection system

Abstract

Stress has now become a ubiquitous part of the fast-moving life, due to which many people are affected. Stress, is identified by physical signs of tension, like irritation, anger, nervousness and sadness at an exceeding level. A stressed individual has an abnormal heart rate, blood pressure and breathing. This may cause major variations in mood, productive lifestyle, and quality of life. This work concentrates on detecting the stress of a person by using the time series analysis of Electromyogram (EMG), Galvanic Skin Response (GSR hand and foot), Electro- cardiogram (ECG) levels collected from physionet database. The obtained data is analysed and a dataset with healthy and stressed population is prepared. This work concentrates on improving the performance of a stress detection system using Support Vector Machine classifier. The Performance of the proposed system is measured using metrics like accuracy, sensitivity and specificity. A significant improvement in the metrics of the proposed system claims that this method will aid in diagnosing the stress rate of a person and aftermath necessary steps required to reduce the stress of the being.