BN PD Risk Model
Parkinson’s Disease (PD) is a complex neurodegenerative disorder marked by patient- specific motor symptoms and non-motor symptoms. Despite of a number of past experimental and modelling efforts there is still a lack of biomarkers and mechanistic understanding of PD disease progression. Longitudinal models of patient trajectories could help closing this gap. A longitudinal Bayesian network (BN) for Parkinson’s disease (PD) was built to model and predict the disease progression using heterogeneous and multi-modal data provided by Parkinson’s Progression Markers Initiative (PPMI). The longitudinal Bayesian network was learned using values of predictors of disease progression at baseline and follow-up clinical visits. The approach also addresses the issue of patient drop out during the study. Moreover, the model is fully generative and can thus potentially be used to simulate virtual patient cohorts and (therapeutic) interventions into such cohorts.