Accounting for Underspecification in Statistical Claims of Model Superiority
This addresses the need for more robust statistical validation in medical imaging systems, though it is incremental as it builds on existing frameworks.
The paper tackled the problem of false superiority claims in machine learning for medical imaging by extending a statistical framework to include underspecification, showing that even modest seed variability (~1%) significantly increases the evidence needed to support such claims.
Machine learning methods are increasingly applied in medical imaging, yet many reported improvements lack statistical robustness: recent works have highlighted that small but significant performance gains are highly likely to be false positives. However, these analyses do not take \emph{underspecification} into account -- the fact that models achieving similar validation scores may behave differently on unseen data due to random initialization or training dynamics. Here, we extend a recent statistical framework modeling false outperformance claims to include underspecification as an additional variance component. Our simulations demonstrate that even modest seed variability ($\sim1\%$) substantially increases the evidence required to support superiority claims. Our findings underscore the need for explicit modeling of training variance when validating medical imaging systems.