Monitoring of Physiological Response Specific to Sleep Deprivation

Sleep deprivation (SD) is a common phenomenon that has attracted increasing attention in society due to its potential impact on health and work efficiency. This study proposes a machine learning-based approach to detect physiological responses specific to SD (PRSSD), aiming to identify PRSSD in the daily lives of college students to promote better health and improve work performance. The experimental data was obtained from 424 samples of 129 college students, and their electrocardiogram and three-dimensional acceleration signals were collected from 13:40 to 14:00, extracting a total of 35 physiological features. Five binary classification models were trained using these features, and key feature subsets were identified using a leave-one-subject-out cross test method and forward feature selection algorithm to construct a PRSSD detection model for short-term monitoring of PRSSD. Among the five models trained, the GradientBoosting Classifier showed the best performance, with an F1_score of 73.09 %.