For the first objective, digital gait and mobility data from the LuxPARK and LWSD cohorts will be utilized. Statistical methods, AI/ML and XAI techniques will be applied to detect the most relevant sensor-derived gait features and relative test paradigms for the assessment of the different mobility impairment domains of PD (e.g., bradykinesia, balance, mobility, freezing). The clinical scores collected by clinicians during the visit will be considered as ground truth and, as such, representative of the patient’s condition. Once the most relevant features have been identified, their clinical utility will be validated, demonstrating their sensitivity to change after physiotherapy and measuring their impact on HCP decision. Physiotherapists will tailor the treatment of the control arm according to standard physical examination carried out at the beginning of the visits, whereas for the interventional arm they will also have access to sensor-derived gait information recorded in the previous days. The longitudinal changes in digital biomarkers will be correlated with the clinical global impression from physiotherapists and PROMs, to determine the feasibility of using sensor-derived information as markers of treatment response. The difference between arms will evaluate the impact of objective sensor measurements in terms of clinical decision support. DC12 will work closely together with DC4 (digital speech) with regard to the question of treatment response monitoring.
This project is part of the "Digital Health" work package.
Artificial Intelligence in Parkinson’s Disease