Sleep Quality through Vocal Analysis: a Telemedicine Application
Federica Amato, Irene Rechichi, Luigi Borzì, Gabriella Olmo
Abstract
Voice is a reservoir of valuable health data. Recent studies highlighted its efficacy in predicting sleep quality, and its potential as biomarker of neurodegeneration. This study assesses the feasibility of a Telemedicine system for the evaluation of sleep quality through brief vocal recordings. Machine Learning models were employed in the binary classification between good and poor sleepers, with great performance in scoring poor sleep quality – 88% and 85% F-1 score on a 5-fold Cross Validation (CV) for females and males, respectively. Moreover, the correlation between perceived sleep quality and a validated global score was studied, as well as the influence of external factors and sleep-wake schedule.