Objective: To model the progression of PD by developing a hidden Markov model which incorporates nonlinear effects of disease severity and medication on clinical scores.

Background: Parkinson’s disease (PD) is a heterogeneous disorder both clinically and pathophysiologically. Accurate ways to group individuals with PD according to common characteristics and rates of progression are critical to both clinical decision-making and trial design.

Method: The model was learned and applied using 10-years of longitudinal clinical data from the Parkinsons Progression Markers Initiative (PPMI) (n=417 PD subjects who were newly diagnosed and untreated at enrollment). A range of clinician- and patient-reported rating scales for motor and non-motor features were included in the model, totaling 78 items. The personalized medication effects model allowed for both improvement in PD symptoms and signs as well as adverse effects of medications, which were learned using the data.

Results: The model has discovered distinct disease states each of which has associated clinical presentation and progression patterns, including occurrence of key outcomes such as dementia and motor complications. Validation of the learned states includes biological characterization by comparing candidate biofluid biomarkers of neurodegeneration and dopamine transporter imaging.

Conclusion: We developed a statistical progression model of Parkinson’s disease that accounts for medication effects. Open access to this model will enhance replication and reproducibility of progression modeling in the field.

« Back to 2023 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/an-open-access-model-for-parkinsons-disease-progression/