Smartphone as unobtrusive sensor for real-time sleep recognition

Sleep is fundamental to health, performance and well-being. Studies demonstrate that, in some countries, sleep disorders are reaching epidemic levels. For this reason, automatic sleep recognition systems can be helpful, on the one hand, to foster self awareness of own habits and, on the other, to implement environment management policies to encourage sleep. In this context, we propose an unobtrusive smartphone application which relies on contextual and usage information to infer sleep habits in real-time. We test selected features using kNearest Neighbors, Decision Tree, Random Forest, and Support Vector Machine classifiers. Moreover, we exploit a 1st-order Markov Chain to improve the algorithm's performance. Experimental results demonstrate the effectiveness of the proposed approach, achieving acceptable results in term of Precision, Recall, and F1-score.