Importance of the time alignment of the sleep probabilistic curves

Sleep is a dynamic process which can be described by a finite set of sleep stages. In contrast to the standard discrete Rechtschaffen and Kales sleep model continuous sleep representation provided by the Probabilistic Sleep Model allows to take into account the whole dynamic of the sleep process. However, analysis of the sleep probabilistic curves faces problems when the time misalignment is present. In this article we highlight the necessity of curves synchronization, or time alignment, before further analysis. In time aligned and original sleep probabilistic curves are transformed into a finite dimensional vector space which is used to predict age or daily measures. Using this approach we observe that curves alignment significantly improves the prediction of the daily measures especially in the case of the REM sleep stage.