Study Rationale: We used to think of Parkinson's disease (PD) as a single disorder. However, individuals with PD can be very different in terms of their clinical characteristics and the biological processes that underlie their disease. We aim to apply artificial intelligence to datasets collected from large patient cohorts to propose distinct subtypes of the disease.
Hypothesis: We hypothesize that our analyses will allow us to identify subtypes that will differentiate individuals with PD based on clinical and biological information.
Study Design: We will begin by standardizing the data contained in two publicly available PD datasets. Next, we will run machine-learning models to generate PD subtypes. Finally, we will determine whether these subtypes accurately distinguish individuals in terms of their biomarker profiles, genetic contributions and clinical presentation.
Impact on Diagnosis/Treatment of Parkinson’s disease: We hope to help in developing new treatments for PD.
Next Steps for Development: If our approach is successful, it can be used to better select individuals for clinical trials and to determine the most effective treatment protocols.