Researchgpt 4clinical diagnosticsfabp4
Researchers Pair AI With Biomarker To Diagnose Pneumonia
9.0
UC San Francisco researchers published Dec. 16 in Nature Communications an observational study pairing GPT-4 analysis of medical records with an FABP4 biomarker to diagnose lower respiratory infections in critically ill adults. In 157 patients (98 pre-COVID, 59 during COVID), the combined approach reached 96% diagnostic accuracy and could have cut inappropriate antibiotic use by more than 80%, speeding diagnosis versus culture.
Key Points
- 1Show combined GPT-4 analysis and FABP4 biomarker achieved 96% diagnostic accuracy in critically ill patients
- 2Show it distinguished infectious from non-infectious respiratory failure more accurately than ICU clinicians
- 3Indicate potential to reduce inappropriate antibiotic use by over 80%, enabling faster, targeted therapy
Scoring Rationale
High clinical novelty and peer-reviewed validation, but limited cohort size and early-stage clinical deployment constrain immediate generalizability.
Sources
Public references used for this report.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
Used by DS/ML engineers at top companies
250 free problems · No credit card
See all Health & Insurance problems