HealthMore in Health→
AI Health Prediction Models Built on Unverifiable Datasets Raise Serious Concerns
Key Takeaways
- AI models for stroke and diabetes risk may be based on datasets with unverifiable origins.
- Two Kaggle datasets scored 0/9 on data-provenance criteria.
- Datasets were used in 125 peer-reviewed studies and cited in a medical device patent.
- Authors call for stronger disclosure requirements and removal of unreliable datasets.
DE
DT Editorial Team··via medicalxpress.com