About

The analysis of polysomnography (PSG), the standard test for the diagnosis of sleep disorders, constitutes one of the most time-consuming tasks in the daily work of a sleep center.

A typical PSG examination contains somewhere between six up to twenty-four hours of continuous neurophysiological activity recording, subject to subsequent visual scoring by clinical experts. Furthermore, a clinician’s time is expensive and scant, and the large amount of data and their complexity makes PSG analysis a task prone to errors and to subjective interpretations.

The use of semi or automatic PSG scoring presents potential advantages. First, by supporting clinicians in the analysis, great savings in scoring time can be achieved, reducing diagnostic costs. In addition, has the interesting property of providing deterministic (repeatable) outcomes. However, despite several early attempts and abundant scientific literature, the problem remains unsolved, and practical acceptance of automatic PSG analysis among the clinical community remains low.

The main objective of this project is to develop an intelligent, automated system that supports the analysis of sleep study data, intending to make the process more accurate, efficient, and accessible.
The project focuses on three core areas:
Use attention-based deep learning models, focusing on the main limitations identified in current first-generation state-of-the-art approaches.
Integrate expert knowledge (Human-in-the-loop approaches) to help these deep learning models learn more effectively with less data.
Use of quantum computing and quantum machine learning to better handle high-dimensional data and complex correlations to improve the performance of the models.

We invite you to dive into the NEXT-GEN-SOMNUS project and discover our innovative research, along with the precise methodologies we're actively implementing.

If our scientific work aligns with your goals, please don't hesitate to connect with us.