On October 17, 2023, Professor , Dean of the Faculty of Informatics at the Università della Svizzera italiana (USI) in Lugano, Switzerland gave a lecture on the Horizon project “SmartCHANGE” and privacy and trust in the context of smart health assistants. Before the lecture, RTG’s scientific coordinator, Dr. Ephraim Zimmer presented RTG Privacy and Trust for Mobile users, and after the lecture our doctoral candidates engaged in an interesting discussion with the guest. Marc Langheinrich
The Horizon Europe project “SmartCHANGE” (2023-2026) aims to develop advanced AI-powered models capable of predicting long-term risks of cardiovascular and metabolic diseases in children. Apart from developing AI models that can use the available physiological data to predict one's future health many years into the future, the project also needs to address the two elephants in the room, i.e. privacy and trust. It necessary to make sure that physiological data from children will not be shared in order to train our AI models, while we must ensure that end-users will sufficiently trust the recommendation of the system in order to actually adopt any suggested lifestyle changes.
The researchers at USI works on both of these aspects: (1) As part of their work in Explainable AI (XAI), they are creating system modules that can explain end-users the reasoning behind the predictions and provide actionable suggestion (“counterfactuals”) to change these predicted outcomes; while (2) to ensure that such explanations and suggestions are done in a privacy-aware manner, they are devising a Federated Learning based approach for finding these counterfactuals.