A study criticized for data leakage used anti-hypertensive-drug use as a feature to predict hypertension, even though that information would not be available when predicting the outcome for a new patient.
Notes on verification
Confirmed by the original 2021 JMIR comment (Chiavegatto Filho et al.) critiquing Ye et al. 2018, plus independent citation in a 2024 arXiv preprocessing pitfalls paper and the Kapoor & Narayanan review. Multiple independent sources corroborate the specific mechanism of leakage. [tier=gold indep_score=0.983 clusters=3 claim_tier=notable]
Sources
- Leakage and the reproducibility crisis in machine-learning-based science (seed:psychology_and_neuroscience)
- https://pmc.ncbi.nlm.nih.gov/articles/PMC7880048/ (corroboration)
- https://www.jmir.org/2021/2/e10969 (corroboration)
- https://arxiv.org/pdf/2412.03491 (corroboration)
- https://www.jmir.org/2018/1/e22/ (corroboration)