Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
June 14, 2026International Journal of Medical InformaticsOpen Access

Why almost all ML models for medicine are wrong-and what we need for evidence-based medical AI

View Full Paper
Ask AI
Bookmark
Share

Authors

FCFederico CabitzaGJGiuseppe JurmanFMFilippo Molinari

Discussion

Loading...

Member takes

Overview

Editorial highlights methodological issues in ML for medicine, suggesting improvements for credible clinical applications.

Key Points

  • This editorial addresses weaknesses in the evidentiary basis of machine learning models in medicine. It aims to improve their reliability in clinical settings.
  • Analyzes current medical ML pipelines for methodological shortcomings.
  • Proposes standards for evidence-based medical AI, including better annotation practices and uncertainty modeling.
  • Calls for rigorous external validation and post-deployment monitoring.
  • Identifies multiple weaknesses in existing medical ML models such as reliance on uncertain ground truths and unstable metrics.
  • Suggests that current practices produce optimistic performance estimates, impacting their clinical utility.
  • Emphasizes the necessity of stricter standards for predictive models in medicine.

Cite This Study

Cabitza et al. (2026) studied this question.

synapsesocial.com/papers/6a2e456cb1cc60ccdea8a7c6https://doi.org/10.1016/j.ijmedinf.2026.106538
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1AI Cautionary Guide: Pitfalls and Strategies for the Use of Machine Learning in Medical Research2026
  2. 2A Scoping Review of AI/ML Algorithm Updating Practices for Model Continuity and Patient Safety Using a Simplified Checklist.2025
  3. 3Evidence over explanations: put medical AI to the test2026 · 1 citations
  4. 4Harnessing Artificial Intelligence and Advanced Machine Learning Algorithms for Medical Data Discovery, Storage, Analysis, and Interpretation: Implications for Physicians, Disease Management, and Patient Health Outcomes2026
  5. 5A responsible artificial intelligence framework for translational readiness of machine learning biomarker models in clinical decision support2026