Use of Deep Learning Techniques for Motor Events Detection in Polysomnographic Records

Abstract

[Abstract] Sleep medicine deals with the diagnosis and treatment of sleep-related disorders. The diagnosis is carried out through the manual analysis and labeling of polysomnographic studies, which record various electrophysiological and pneumological signals of the patient throughout the night. This process involves the analysis of long duration signals, is complex, and demands considerable resources and time on the part of the clinical expert. The purpose of this proyect is the construction of automatic analysis algorithms that considerably reduce the analysis duration, reducing the manual workload, and minimizing possible human errors, providing repeatability and robustness. In particular, the objective is to use machine learning algorithms, based on Deep Learning techniques, for the identification and location of physiological events in these polysomnographic records. Specifically, the goal is to locate physiological events associated with involuntary motor movements that occur in the limbs, known as Limb Movements