WCN25-1426 EVALUATING CHATGPT-4'S ACCURACY IN IDENTIFYING NEPHROLOGY MEDICATIONS FROM SELF-CAPTURED PILL IMAGES ACROSS MULTIPLE GPT-4 MODELS
sodium, osmolarity, glucose, TSH, cortisol, BUN, creatinine, urine osmolarity, urine sodium, and volume status.. Iso-osmolar scenarios were assessed both with and without lipid and free light chain levels.Data Collection: GPT-4o mini's responses were evaluated with and without predefined answer options provided for all cases on two separate occasions, 24 hours apart.The answer options included: A. Hyperlipidemia or Hyperparaproteinemia B. Hyperglycemia C. Mannitol use D. Diarrhea or prior diuretic use E. SIADH F. Current diuretic use with a thiazide diuretic G.
Authors
- Wisit Cheungpasitporn (ORCID: https://orcid.org/0000-0001-9954-9711)
- Iasmina Craici
- Jing Miao (ORCID: https://orcid.org/0000-0003-0642-9740)
- Erin F. Barreto (ORCID: https://orcid.org/0000-0002-0996-1487)
- Fawad Qureshi (ORCID: https://orcid.org/0000-0003-3387-4882)
- Supawadee Suppadungsuk (ORCID: https://orcid.org/0000-0003-1597-2411)
- Benjamin Dreesman (ORCID: https://orcid.org/0009-0006-7142-710X)
- Charat Thongprayoon (ORCID: https://orcid.org/0000-0002-8313-3604)
- Kianoush Kashani (ORCID: https://orcid.org/0000-0003-2184-3683)
- Michael A. Mao (ORCID: https://orcid.org/0000-0003-1814-7003)
- M. Salman Sheikh
Institutions
- Mayo Clinic (US)
- Jacksonville College (US)
- Mayo Clinic in Arizona (US)
- Hypertension Institute (US)
- Mayo Clinic in Florida (US)
Publication Details
- Journal
- Kidney International Reports
- Published
- 2025-01-27
- DOI
- https://doi.org/10.1016/j.ekir.2024.11.176
- Primary Topic
- Artificial Intelligence in Healthcare
- Type
- article
- Field-Weighted Citation Impact
- 0.00