• The CITIC of the UDC will collaborate with the Galician Supercomputing Center (CESGA) to maximize the advantages of its NEXT-GEN-SOMNUS project, aimed at developing more efficient ICT solutions for studying sleep disorders, which currently affect more than four million people in Spain.

A Coruña, February 27, 2025.– CITIC continues to strengthen alliances in the development of projects that apply artificial intelligence and quantum computing to different areas of daily life. Specifically, the UDC research center has established synergies with the Galician Supercomputing Center (CESGA) to collaborate on the project “Next-Generation Machine Learning Algorithms for the Analysis of Sleep Medical Records (NEXT-GEN-SOMNUS)”. The project aims to develop more efficient ICT solutions for studying sleep disorders, a problem that currently affects more than four million people in Spain, according to data from the Spanish Society of Neurology.

The initiative, coordinated by the teams of Eduardo Mosqueira Rey and Diego Álvarez Estévez at CITIC under the ‘2023 Knowledge Generation Projects’ call, is funded by the Ministry of Science, Innovation and Universities, the State Research Agency, and the European Regional Development Fund (ERDF). It is designed to improve the analysis of sleep medical records through the use of next-generation machine learning algorithms based on self-attention mechanisms, expert human collaboration in the learning process, and the inclusion of quantum computing. Through these approaches, Mosqueira explains, “the goal is to develop more efficient and explainable solutions for the diagnosis of sleep disorders”.

The collaboration with CESGA, the researcher adds, “involves using CESGA’s quantum time series prediction model, adapting it for the type of signals we use in sleep medicine, and properly adjusting it to perform sleep phase classification tasks and detect various events relevant to diagnosis. The adapted model would be tested with real-world sleep medicine data and evaluated both in quantum simulators and the actual quantum computer installed at CESGA”.

Accelerating Diagnosis for Greater Clinical Efficiency

Sleep disorders affect a significant portion of the population. According to statistics from the Spanish Society of Neurology, more than four million people in Spain—48% of the adult population and 25% of children—do not experience quality sleep. The diagnostic procedures associated with manually reviewing standard polysomnography (PSG) tests are complex and costly, making it difficult for clinical centers to meet the growing demand for these examinations.

Computer-assisted PSG analysis offers clear advantages in terms of significant time savings and reduced overall diagnostic costs. However, existing solutions are limited to partial, ad hoc implementations and suffer from generalization issues. First-generation deep learning techniques, such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs), are suitable for detecting and classifying single objects (e.g., images) but are not optimal for more complex situations requiring the identification of multiple individual elements. Additionally, these models function as “black boxes,” which negatively impacts their acceptability by clinicians, hindering the responsible use of their decisions.

Against this backdrop, the NEXT-GEN-SOMNUS project at CITIC-UDC aims to explore the applicability of next-generation machine learning techniques to the analysis of sleep medical records, overcoming the limitations of first-generation models. As Mosqueira Rey explains, the project proposes “improving algorithm efficiency by integrating novel self-attention mechanisms to achieve better event detection and classification results. Additionally, incorporating human-in-the-loop techniques to introduce expert human knowledge into these algorithms, enhancing both performance and explainability. Moreover, integrating quantum machine learning tasks that leverage processes such as superposition, interference, and entanglement, taking machine learning algorithms to a new level and maximizing the benefits of these ICT tools for diagnosing, studying, and monitoring sleep disorders”.