To test this hypothesis, we will employ recent discoveries in computer science, PD biomarker research and standardized sets of PD data. First, we will develop machine-learning techniques to analyze multimodal data from people with and without Parkinson's. Results will help identify whether someone does or does not have PD, and they will be grouped accordingly. Second, we will confirm our findings using existing data from Massachusetts General Hospital and the Parkinson's Progression Markers Initiative (PPMI) repository. (PPMI is The Michael J. Fox Foundation's landmark study to find Parkinson's biomarkers.)
Impact on Diagnosis/Treatment of Parkinson's disease:
If successful, this diagnostic approach will help identify clinically relevant and economically viable PD biomarkers. It also can aid primary care physicians and neurologists in clinical decision-making. Finally, it could accelerate drug discovery by helping analyze outcomes of drug trials.
Next Steps for Development:
This project will be followed by a larger, high-powered, multi-institutional PD biomarker study to confirm the value of our diagnostic approach. In the long term, we will expand this approach by developing methods for PD classification based on disease features and prognosis. These methods eventually will support the design of personalized therapeutic regimens for PD.