Researchers at Weill Cornell Medicine have made significant strides in understanding Parkinson’s disease using machine learning techniques, resulting in the identification of three distinct subtypes based on disease progression rates. Published in npj Digital Medicine on July 10, their findings not only promise to enhance diagnostic and prognostic capabilities but also suggest targeted therapeutic avenues based on subtype-specific genetic markers.
Dr. Fei Wang, senior author and director of the Institute of AI for Digital Health at Weill Cornell Medicine, emphasized the heterogeneous nature of Parkinson’s disease, underscoring the need for personalized treatment strategies tailored to each subtype. The study categorized patients into the Inching Pace (PD-I), Moderate Pace (PD-M), and Rapid Pace (PD-R) subtypes, each characterized by varying speeds of symptom advancement.
Using deep learning methodologies, the researchers analyzed extensive clinical data from multiple databases to define these subtypes. They also delved into molecular mechanisms through genetic and transcriptomic analyses, uncovering distinct biological pathways associated with each subtype, such as neuroinflammation and oxidative stress.
Building on their earlier work in the Parkinson’s Progression Markers Initiative (PPMI), the team validated their findings across cohorts from the National Institute of Neurological Disorders and Stroke (NINDS) Parkinson’s Disease Biomarkers Program (PDBP). This comprehensive approach not only confirmed the existence of these subtypes but also identified potential drug candidates that could modify disease progression, including repurposing existing medications like metformin.
Dr. Chang Su, first author of the study, highlighted the beneficial effects of metformin on cognitive symptoms in PD-R subtype patients, suggesting its potential as a therapeutic agent. The research team further underscored the importance of leveraging diverse datasets, such as those from the INSIGHT Clinical Research Network and OneFlorida+ Clinical Research Consortium, to validate their findings in real-world settings.
Collaborators from various institutions contributed to this multidisciplinary effort, reflecting a broad scientific partnership aimed at advancing Parkinson’s research through innovative computational and experimental approaches.
In conclusion, this study not only defines novel Parkinson’s disease subtypes but also paves the way for precision medicine approaches that could revolutionize treatment strategies and improve patient outcomes globally.
Article written by Well Cornell
16/07/2024
Source:
Weill Cornell
https://news.weill.cornell.edu/news/2024/07/machine-learning-helps-define-new-subtypes-of-parkinson%E2%80%99s-disease