arXiv · 2611.14533
Quantifying Methodological Inflation in Clinical Machine Learning: A Cross-Modality Audit of Subject-Level Leakage and Nested Feature Selection for Parkinson's Disease
Abstract
Reported accuracy in Parkinson's disease (PD) machine learning can reflect the evaluation protocol rather than real signal. We ask how much evaluation choices inflate performance by comparing naive and rigorous protocols on voice, telemonitoring and blood-transcriptomic datasets. Subject-aware splitting reduced voice AUROC from 0.892 to 0.780 and telemonitoring R^2 from 0.105 to -0.037; Random Forest recovered neither gap. Supervised probe selection before splitting inflated transcriptomic AUROC from 0.711 to 0.998. A nested 50-probe panel matched a 1,000-probe panel (0.811 versus 0.844), and recruitment-wave batch structure was detected. Future work should validate a stable panel on an independent cohort.
Explore related subjects
Keep this discovery
Explore connections, maps & timelines
Namrata Datta, Victor Miranda Soberanis. 2026-10-09. Quantifying Methodological Inflation in Clinical Machine Learning: A Cross-Modality Audit of Subject-Level Leakage and Nested Feature Selection for Parkinson's Disease. https://arxiv.science/abs/2611.14533
Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite arXiv Science by Indiciae: BibTeX · RIS.