Sleep Quality and Best Posture Prediction using Contextual Body Sensors using LSTM

Sleep quality is one of the most important features to predict a healthy living style. It relates to various physical and physiological diseases. Correct or most comfortable posture is responsible for having and predicting the quality of sleep anyone had. However, posture in which a person is sleeping is his/her personal requirement or desire. Correct posture is context sensitive information which needs to be consider when predicting sleep quality of any person. In this paper a RNN based framework is proposed which predicts posture and compare it with recommended posture of a person. If the person is sleeping in wrong posture, it generates an alert to correct the posture. The proposed framework is tested on standard dataset of sleep and pressure mat available at physionet.org. The proposed framework has an accuracy of above 90% in validation dataset.