AI-PROGNOSIS took part in the EUSIPCO 2026 conference, held from 31 August to 4 September 2026 in Bruges, Belgium, where the project organised the special session “Digital Biomarkers for Chronic Diseases”, co-chaired by Christos Chatzichristos from KU Leuven.

Two AI-PROGNOSIS works were presented during the session.

The first study, “Assessment of Fine Motor Skills Impairment in Parkinson's Disease Using Touchscreen Typing Dynamics: a Multi-Cohort Analysis” by Gerasimou et al., leveraged the timing of key-tap events during touchscreen typing to compute a digital score that strongly correlates with a clinical measure of finger movement slowness in people with Parkinson's disease (PD). Slowness of movement, or bradykinesia, is one of the cardinal motor symptoms of PD. The team presented early evidence of the score's generalisation and test-retest reliability in a sub-cohort of the AI-PROGNOSIS dBM-DEV study (NCT06444789), where participants contributed free-living, routine typing data.

The second study, “Permutation-Invariant Multi-Night Actigraphy for the Screening of REM Sleep Behavior Disorder” by Wang et al., leveraged free-living wrist-acceleration data collected over multiple nights in the AI-PROGNOSIS dBM-DEV study (NCT06444789) to develop a machine learning model capable of distinguishing people with REM sleep behaviour disorder (RBD) from unaffected individuals. RBD is a sleep condition characterised by dream enactment and an early sign of Parkinson's disease. The team compared different machine-learning classifiers and observation windows, while also exploring the movement characteristics during sleep that may serve as proxies of RBD.

AI-PROGNOSIS team members Stelios Hadjidimitriou and Ioannis Gerasimou from AUTH, and Christos Chatzichristos, Fan Wang and Maarten De Vos from KU Leuven attended the conference.