WP1: Data preparation and validation framework

In this WP we will perform the basic tasks that will allow us to start building the ML models. Specifically, we will focus on obtaining the annotated PSG datasets from different sources (where usage of the open EDF standard will be key for simple and flexible data exchange and storage from different data sources). This will include also the typical preprocessing tasks needed to adapt the raw data into a format that is valid for feeding a ML model (integration, cleaning, transformation).

Also, since much of these tasks will be done with the collaborating medical institutions, we will take advantage of the close contact with the physicians to establish and characterize the inter-database differences and establish a baseline of human agreement. Finally, we will include a state-of-the-art analysis of all the three next-generation ML techniques that we want to include in the project in relation to the datasets obtained.