Fast Approximate Energy Minimization via Graph Cuts
IEEE Transactions on Pattern Analysis and Machine Intelligence
| Hi-index | 0.00 |
In this paper a definition of the activity (ACT) variable is proposed and a method to estimate it from the noisy actigraph output sensor data is described. A statistical model for the actigraph data generation process is suggested based on its working physical principles and on physiological considerations about human activity. The purposeless nature of the sleeping movements is used to discriminate the Sleep and Wakefulness (SW) states. The estimated ACT signal from the actigraph output signal is correlated with the data from a Sleep Diary to validate the SW oscillations, computed from the ACT. A Sleep electronic Diary (SeD) was implemented in the scope of this work to make it possible an accurate register of the patient activities relevant for the diagnosis of sleep disorders. Examples using real data, illustrating the application of the method, have shown high correlation between the output of the proposed algorithm that characterizes the activity and the data registered in the SeD.