https://screeninglibraries.com..../venous-thromboembol
Interpretable deep learning models, employing multi-task learning, were trained to estimate FOG (cross-validated F1 score: 976%), medication state (OFF vs. ON levodopa; cross-validated F1 score: 968%), and overall PD severity (MDS-UPDRS-III score prediction error: 27 points). These models used kinematic data from 57 patients during levodopa challenge tests. The proposed model effectively