NPR News30.04.2026ScienceUnited States
AI Diagnostic Model Matches or Outperforms Physicians in Clinical Cases, Study Finds
Research from Harvard and Beth Israel Deaconess shows AI can accurately diagnose patients using electronic health records, though authors caution against replacing doctors
L'essentiel
- Researchers at Harvard Medical School and Beth Israel Deaconess Medical Center found that an OpenAI reasoning model matched or outperformed experienced physicians in diagnosing patients using only electronic health records.
- In one case, the AI correctly identified lupus-related heart inflammation as the underlying cause of a pulmonary embolism when doctors initially suspected medication failure.
- The study published in Science highlights AI's potential to assist clinical diagnosis while emphasizing the technology should augment, not replace, physicians.
Résumé généré par IA
Taille de police
Researchers at Harvard Medical School and Beth Israel Deaconess Medical Center found that an OpenAI reasoning model matched or outperformed experienced physicians in diagnosing patients using only electronic health records. In one case, the AI correctly identified lupus-related heart inflammation as the underlying cause of a pulmonary embolism when doctors initially suspected medication failure. The study published in Science highlights AI's potential to assist clinical diagnosis while emphasizing the technology should augment, not replace, physicians.
Sujets liés
This article was originally published by NPR News.
