Ricerc@Sapienza

The remote monitoring and management of Parkinson's Disease (PD) patients is still a challenge. Nowadays, there is no operationally and economically feasible system that can truly support PD patients at home. Wearable sensor systems and camera-based systems for health monitoring are an emerging trend and are expected to enable proactive personal health management. Long-term physical activity monitoring may provide valid information about disease progression, rehabilitation success, and effects of medical and surgical interventions in PD patients.
By using ad-hoc machine learning tools and hardware technologies, the monitoring of medical treatment will be improved, thus obtaining a better control with reduction of motor fluctuations, as well as quantify the risk of falls in patients suffering from PD. The goal of avoiding hospitalization and replace it with home monitoring is relevant because patients would increase their well-being with an optimized therapy, avoid hospitalization and keep their lifestyle. Actually, the National Health System would save a relevant amount of economic resources with relevant socio-economic advantages.
This project aims at developing an integrated data processing system that, on the basis of different biological/mechanical signals collected from the body and from the environment, will implement a decision support system for suited diagnostic or preventive actions. This will be obtained by using in a novel fashion different and complementary sensor technologies and suited data fusion procedures. A machine learning approach will be adopted for which the quality model is learnt from ground truth data, which represent activity samples at the best affordable quality. Although the use of a body area network of wearable sensing unit enables to investigate the dynamics of all the different body segments, this solution is not really unobtrusive for the user and it is difficult to implement such an approach in unsupervised settings.