Detecting Sleep Disorders in Polysomnography Data

The primary diagnostic test for sleep disorders is known as polysomnography, traditionally conducted while a patient sleeps under observation in a specialized clinic. The evolution of the Internet of Things (IoT) field has facilitated the collection of various biometric signals, which are also monitored during this examination through wearable devices. In this context, the objective of this work is to explore the utilization of machine learning algorithms in sleep disorder detection. Classification algorithms were assessed, and their outcomes were compared by reducing the number of input signals, aiming to enable the application of polysomnography in a home setting. The results indicate that the most promising algorithm is the Random Forest, which demonstrates satisfactory performance even with a reduced number of input data.