Leveraging Multi-Omics and Clinical Datasets of Parkinson's Disease with Machine Learning

In this thesis, I leverage the wealth of blood transcriptomic, CSF proteomics, and clinical data, including UPDRS and UPSIT scores, meticulously rening the data quality through thorough preprocessing. Employing a progressive feature selection technique, I pinpoint the most crucial genes, and proteins associated with Parkin- son's disease. Subsequently, I deploy a boosting algorithm to construct a diagnostic framework centered around these identied genes and proteins. Additionally, I con- duct an in-depth analysis of UPDRS and UPSIT datasets from PPMI, providing a comprehensive comparison. This holistic approach facilitates a more robust un- derstanding of Parkinson's disease, oering insights for enhanced diagnostic and treatment strategies.