System Maintenance Notice

Please note that maintenance work is scheduled between 4 AM and 5 AM - CET. During this time, you may experience reduced performance. Thanks for your patience.

Publications

Journal Article Ahead of print

PubMed: 42533626 DOI: 10.1093/jamia/ocag127

Tipping the balance: impact of class imbalance correction on the performance of clinical risk prediction models.

J Am Med Inform Assoc

Authors: Amalie Koch Andersen, Hadi Mehdizavareh, Arijit Khan, Tobias Becher, Simone Britsch, Markward Britsch, Morten Bøttcher, Simon Winther, Palle Duun Rohde, Morten Hasselstrøm Jensen, Simon Lebech Cichosz

Affiliations

  • Medical Informatics and Image Analysis, Department of Health Science and Technology, Aalborg University, Aalborg, 9260, Denmark. Medical and Translational Science, Novo Nordisk A/S, Søborg, 2860, Denmark.
  • Department of Computer Science, Aalborg University, Aalborg, 9220, Denmark.
  • Department of Computer Science, Aalborg University, Aalborg, 9220, Denmark. Department of Computer Science, Bowling Green State University, Bowling Green, OH, 43403, United States.
  • Cardiology, Angiology, Haemostaseology, and Medical Intensive Care, Medical Centre Mannheim, Medical Faculty Mannheim, Heidelberg University, Heidelberg, 69117, Germany.
  • European Centre for AngioScience (ECAS), German Centre for Cardiovascular Research (DZHK) partner site Heidelberg/Mannheim, and Centre for Cardiovascular Acute Medicine Mannheim (ZKAM), Medical Centre Mannheim, Medical Faculty Mannheim, Heidelberg University, Heidelberg, 69117, Germany.
  • HMS Analytical Software GmbH, Heidelberg, 69117, Germany.
  • Department of Radiation Oncology, Medical Center Mannheim, Medical Faculty Mannheim, Heidelberg University, Heidelberg, 69117, Germany.
  • DKFZ Hector Cancer Institute, Medical Centre Mannheim, Medical Faculty Mannheim, Heidelberg University, Heidelberg, 69117, Germany.
  • Mannheim Institute for Intelligent Systems in Medicine (MIISM), Heidelberg University, Heidelberg, 69117, Germany.
  • Department of Cardiology, Gødstrup Hospital, Herning 7400, Denmark.
  • Department of Clinical Medicine, Aarhus University, Aarhus N 8200, Denmark.
  • Genomic Medicine, Department of Health Science and Technology, Aalborg University, Aalborg, 9260, Denmark.
  • Medical Informatics and Image Analysis, Department of Health Science and Technology, Aalborg University, Aalborg, 9260, Denmark.
  • Medical and Translational Science, Novo Nordisk A/S, Søborg, 2860, Denmark.

Abstract

OBJECTIVES

Machine-learning-based clinical risk prediction models are increasingly used to support decision-making in healthcare. While class-imbalance correction techniques are commonly applied to address rare outcomes, their impact on probabilistic calibration remains insufficiently understood. This study evaluated the effect of widely used resampling strategies on both discrimination and calibration across real-world clinical prediction tasks.

MATERIALS AND METHODS

Ten clinical datasets spanning diverse medical domains and including over 600 000 patients were analyzed. Multiple machine-learning model families were evaluated. Models were trained on original data and using 3 1:1 class-imbalance correction strategies (synthetic minority oversampling technique, random undersampling, and random oversampling). Performance was assessed on held-out data using discrimination and calibration metrics.

RESULTS

Resampling had no positive impact on predictive performance. Changes in area under the receiver operating characteristic curve (ROC-AUC) and precision-recall AUC were small and inconsistent (ROC-AUC: -0.002 to -0.01; PR-AUC: -0.10 to -0.03), with no method showing systematic improvement. In contrast, calibration was consistently degraded. Resampled models showed higher Brier scores (increase 0.029-0.080) and marked deviations in calibration intercept and slope, indicating distorted predicted risks despite preserved ranking performance.

DISCUSSION

Across diverse clinical datasets, resampling primarily altered the implicit class prior learned during training, leading to miscalibration when models were evaluated. The consistent dissociation between discrimination and calibration highlights that rank-based metrics alone are insufficient for evaluating clinical utility. Gains from imbalance correction can typically be reproduced by threshold adjustment without distorting predicted probabilities.

CONCLUSION

Common 1:1 class-imbalance correction techniques do not improve discrimination and may substantially degrade calibration, limiting their suitability for clinical risk prediction where accurate probabilities are essential.

MeSH

Medical Informatics Applications, Medical Informatics, Probability, Learning