Leveraging machine learning to predict Parkinson's disease using pre-symptomatic proteomics data.

Area of Science:

• Neuroscience
• Proteomics
• Biochemistry

Background:

• Non-motor symptoms of Parkinson's disease (PD) can appear 20 years before motor symptoms.
• Predicting PD risk accurately remains a significant clinical challenge.
• Early detection is crucial for timely intervention and disease management.

Purpose of the Study:

• To investigate the potential of plasma proteomic signatures for predicting Parkinson's disease risk.
• To apply machine learning (ML) models to identify predictive protein biomarkers in pre-symptomatic individuals.
• To explore the biological pathways associated with early PD development.

Main Methods:

• Analysis of 2,937 plasma proteins from UK Biobank participants, including incident and prevalent PD cases and controls.
• Application of ML algorithms to predict PD diagnosis up to 14 years prior.
• Validation of identified proteomic signatures in an independent cohort.
• Pathway enrichment analysis and co-expression network analysis.

Main Results:

• 446 plasma proteins were found to be dysregulated in individuals who later developed PD.
• A panel of 23 proteins achieved an Area Under the Curve (AUC) of 0.78 for predicting incident PD.
• Proteomic signatures were validated in an independent cohort with an AUC up to 0.76.
• Enrichment of PD-related pathways, including Neuron Death and Amyloid-beta Clearance, was observed up to 9 years before diagnosis.

Conclusions:

• Plasma proteomic signatures can predict Parkinson's disease risk up to 14 years before clinical diagnosis.
• Machine learning applied to large-scale proteomics data holds significant promise for early PD detection.
• Dysregulated proteins and pathways provide insights into the early molecular mechanisms of Parkinson's disease.