Performance of protein panels is inflated across many biomarker studies
In the authors' words
Data leakage is a prevalent yet underappreciated flaw in biomarker discovery studies. Through simulation and real-world proteomic data, we demonstrate that typical pipelines are broadly susceptible to this issue, producing inflated performance estimates, poor generalization, and excess false positives. We further introduce two tools to detect data leakage at the code and manuscript level, providing a practical path toward more rigorous and reproducible biomarker reporting.
Main resultThe abstract does not state a limitation.
Appeared: Sunday, September 27. medRxiv. Preprint, not yet peer-reviewed.