Stress prediction through sleep parameters using a feature engineering approach

Stress is a high growing problem in today's age. Stress affects physical and emotional well-being and also affects the normal working of a person. It can have long- lasting health hazards, which can turn into diseases if not treated promptly. Detecting Stress and measuring Stress is the first step toward curing it. Stress can be predicted by using sleep parameters. A sleeping person's snoring range, eye movement, limb movement, hours of Sleep, heart rate, etc., can be used to measure stress levels. Stress can be predicted with the help of these parameters by using Machine Learning models. This study is on the detection of Stress and aims to develop a machine-learning model that can precisely determine stress levels by looking at sleep patterns. In this study, the researcher aims to estimate the amount of Stress using machine learning algorithms using particular human body parameters recorded during Sleep. Among the sleep parameters are snoring intensity, limb movement rate, eye movement rate, and sleep duration. Other metrics include heart rate, respiration rate, body temperature, and oxygen level while sleeping. In order to assess the precision of stress level prediction, the study performs multiclass classification using machine learning techniques. In this research, multiple multiclass classification techniques were used to predict the stress levels of a sleeping person with the highest accuracy of 99.5%. The techniques used in performing the predictions are less in time complexity and code complexity and thus can be used in devices with less computational resources, such as smart wearables. This research will therefore help in detecting and measuring Stress quickly.