Università degli Studi di Pavia

The University of Pavia (PV) is an Italian HEI boasting a centuries-old tradition of excellence. At present, it is home to a vibrant and diverse academic community composed by more than 24,000 students, almost 1,000 members of the teaching and research staff and almost 900 administrative and technical resources. According to the main Italian university ranking (CENSIS), PV has confirmed its 1st place among “big universities” (20.000-40.000 enrolled students).

PV has signed more than 400 collaboration agreements for the development of national, EU and international projects with public and private institutions, as well as partner universities all over the world, and is an active member in a number of international networks (such as the Coimbra Group, EUA and Netval).

At present, PV hosts different EU-funded projects, such as 14 MSCA (2 ITN, 5 IF/PF, 3 DN and 4 RISE), 11 ERC grants (7 Starting, 2 Proof of Concept and 2 Advanced), 3 EIC (1 Pathfinder and 2 Transition) and is actively involved in other 33 research and innovation European R&I projects for a total funding exceeding 29,5M€. At the national level, PV participates in 13 initiatives under the National Recovery and Resiliency Plan (PNRR) and has received almost €71M: 3 National Centres, 4 Extended Partnerships (one of them as main Hub), 1 Innovation Ecosystem, 1 Research Infrastructure, 1 Innovative Technological Infrastructure, 2 National Complementary Plans and 1 Program for Historical Parks and Gardens. Finally, at the national and regional level, PV manages public and private funding worth €25.5M: among the most important projects, 70 PRIN, 9 Lombardy Region and 25 Cariplo foundation projects stand out.

Scientific productivity is documented by more than 25.000 scientific papers, 4.500 book chapters, 1.100 PhD thesis and 75 patents in the last 10 years.

Role within MusiC4Diabetes

The main contribution of the PV team will consist in the design of the advanced adaptive personalized multi-target control algorithm (Model Predictive Control, MPC) considering simultaneously Multi-Metabolite (MM) signals to optimize peritoneal insulin delivery without the obtrusive meal and physical exercise announcements. In details the work of PV team will be devoted to the:

• Identification of new input-output models describing the relationship between MM signals and intraperitoneal (IP) insulin delivery via novel ad-hoc machine learning approaches.
• Design of a MPC to handle MM signals with a multi objective cost function to consider them in different experimental situation (e.g., postprandial glucose control, night control, exercise control, etc.). The IP insulin delivery route does not require meal announcement as well as the lactate signal is ideal for physical activity management without announcement.
• Personalization of the control action via novel ad-hoc machine learning approaches to handle the significant metabolic inter-subject variability.
• Adaptation of the control action to deal with the circadian metabolic variability tracking dynamically the time-variant physiological features and adapting the control strategy accordingly.
• Design of safety tools to guarantee anomalies detection guided by the 3β OH-butyrate signal based on self-diagnostic fault-detection algorithms.