Study Rationale:
Parkinson’s disease and Lewy body disease share many clinical and biological features, which makes it difficult to distinguish between them with current diagnostic tools. This lack of precision limits our ability to identify patients correctly and to understand the unique biology of each condition. By using gene activity data in post-mortem brain tissue, we can learn much more about the biological processes behind these diseases. This project uses advanced machine learning methods to uncover which genes and cell types are most involved in Lewy body disease and how these patterns differ from other neurodegenerative disorders.
Hypothesis:
We hypothesize that advanced machine-learning methods applied to clinical, pathology, and gene-expression data will uncover new molecular patterns linked to Lewy body disease. By identifying the features that most strongly drive predictions, these models will improve diagnostic accuracy beyond what clinical information alone can provide.
Study Design:
We will analyze two large collections of human brain data that include RNA-sequencing data, clinical information, and detailed pathological data. Using machine learning, we will build models that estimates whether someone had Lewy body disease and identify which genes and cell types drive these predictions. We will also use single-cell data from Parkinson’s disease to better understand the roles of neurons and glial cells.
Impact on Diagnosis/Treatment of Parkinson’s disease:
This work could lead to more accurate ways to identify Lewy body disease and related conditions, helping doctors understand disease processes. It may also highlight new biological targets that guide future treatments.
Next Steps for Development:
If successful, the next steps would involve testing the identified gene signatures in living patients, developing blood- or imaging-based biomarkers, and refining those with promising predictive performance, for clinical use.