Biomed

Wearable Monitoring

Currently, many disorders like sleep apnea, epilepsy, and sudden cardiac death are diagnosed in a hospital environment, where medical tests are performed to monitor different physiological parameters. These tests are associated with high costs and low comfort because they typically involve an expensive hospital stay in which patients are monitored with several electrodes and bulky, uncomfortable equipment. Moreover, the usability of these medical tests is limited as they require that patients interrupt their daily activities in order to be monitored, and rare events like seizures and sudden cardiac death may not occur during the limited monitoring time. For this reason, it is important to develop methodologies that can be used in a home or ambulatory environment, and that can achieve comparable diagnostic accuracy to the in-hospital tests.

The main goal of our research in this area is the development of new data analysis tools to characterize and interpret the physiological signals that are affected by different disorders. We specifically focus on sleep and epilepsy monitoring. Some active areas of research are multimodal approaches to machine learning, data-centric AI, reducing the number of electrodes that patients need to wear, dealing with artifacts, and transfer learning and domain adaptation methods to reduce the dependency on large labeled datasets and improve generalization. By developing such systems, the aim is to improve the quality of life of patients, as well as the diagnosis, follow-up, and monitoring of different disorders.