Physics Informed Machine Learning-Based Prediction and Reversion of Impaired Fasting Glucose Management (PRAESIIDIUM)

Project goal and tasks.

The aim of the PRAESIIDIUM project (from Latin, where “praesidium” means “protection”) is to develop a tool aimed at providing a real-time prediction of the pre-diabetic risk of an individual. The prediction algorithm will be based on a “physics-informed machine learning” approach: a rich dataset of real-life data, obtained from already existing previous and a new clinical trial with continuous data ingestion through wearable sensors, will be combined with mathematical models and eXplainable AI (XAI) techniques, to overcome the limits of “black-box” ML approaches while improving the prediction performances and reducing the computational time of the risk calculation based on the simulations of the mathematical model. The final algorithm will be implemented in a web-based platform, where medical doctors and patients can inject data from several sources (acquisition form connected sensors and manual insertion) and obtain a real time analysis of the risk to develop the pre-diabetic condition over time.

Tasks to be performed by EDI.

In this project, EDI will develop a prototype of a wrist-worn wearable device with bioimpedance sensing and signal processing capabilities (EDI), as well as the associated data collection and processing software and algorithms for heart-rate detection. This prototype will be suitable for pilot deployments in clinical trials. Our goal is to test the device in real-life conditions, and compare its data with data from other, already validated sensors, to validate the accuracy of the device in real-life conditions, collect feedback about user experiences, as well as to help to integrate the device in the final software/hardware platform of the project.