Data leakage in machine learning studies creep into meta-analytic estimates of predictive performance
Open Access
- 30 October 2025
- journal article
- other
- Published by Springer Nature in Molecular Psychiatry
- Vol. 30 (12) , 6070-6071
Abstract
No abstract availableKeywords
Funding Information
- Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWA.1518.22.017, NWA.1518.22.017)
This publication has 10 references indexed in Scilit:
- REFORMS: Consensus-based Recommendations for Machine-learning-based ScienceScience Advances, 2024
- Leakage and the reproducibility crisis in machine-learning-based sciencePatterns, 2023
- I tried a bunch of things: The dangers of unexpected overfitting in classification of brain dataNeuroscience & Biobehavioral Reviews, 2020
- Establishment of Best Practices for Evidence for PredictionJAMA Psychiatry, 2020
- Clinical judgement, case complexity and symptom scores as predictors of outcome in depression: an exploratory analysisBMC Psychiatry, 2020
- Double Dipping in Machine Learning: Problems and SolutionsBiological Psychiatry: Cognitive Neuroscience and Neuroimaging, 2020
- Ten simple rules for predictive modeling of individual differences in neuroimagingNeuroImage, 2019
- PROBAST: A Tool to Assess Risk of Bias and Applicability of Prediction Model Studies: Explanation and ElaborationAnnals of Internal Medicine, 2019
- Cross-validation failure: Small sample sizes lead to large error barsNeuroImage, 2018
- Conducting Meta-Analyses inRwith themetaforPackageJournal of Statistical Software, 2010