Final assignment for my course on Biomedics Data Science applications, with collaboration of my colleague Sergio Conde.
The main goal was the feature extraction from different temporal series, training a classifier (Random Forest) maximizing precission as a clinical requirement, and developing the explainability of our model using Shapley values. Our work follows the publication 'ISRUC-Sleep: A comprehensive public dataset for sleep researchers'. We used data from 10 healthy subjects, which can be found on subgroup 3 here. We used the specialized library YASA for feature extraction.