Navigating Nephrology's Decline Through a GPT-4 Analysis of Internal Medicine Specialties in the United States: Qualitative Study.

Background: The 2024 Nephrology fellowship match data show the declining interest in nephrology in the United States, with an 11% drop in candidates and a mere 66% (321/488) of positions filled.
Objective: The study aims to discern the factors influencing this trend using ChatGPT, a leading chatbot model, for insights into the comparative appeal of nephrology versus other internal medicine specialties.
Methods: Using the GPT-4 model, the study compared nephrology with 13 other internal medicine specialties, evaluating each on 7 criteria including intellectual complexity, work-life balance, procedural involvement, research opportunities, patient relationships, career demand, and financial compensation. Each criterion was assigned scores from 1 to 10, with the cumulative score determining the ranking. The approach included counteracting potential bias by instructing GPT-4 to favor other specialties over nephrology in reverse scenarios.
Results: GPT-4 ranked nephrology only above sleep medicine. While nephrology scored higher than hospice and palliative medicine, it fell short in key criteria such as work-life balance, patient relationships, and career demand. When examining the percentage of filled positions in the 2024 appointment year match, nephrology's filled rate was 66%, only higher than the 45% (155/348) filled rate of geriatric medicine. Nephrology's score decreased by 4%-14% in 5 criteria including intellectual challenge and complexity, procedural involvement, career opportunity and demand, research and academic opportunities, and financial compensation.
Conclusions: ChatGPT does not favor nephrology over most internal medicine specialties, highlighting its diminishing appeal as a career choice. This trend raises significant concerns, especially considering the overall physician shortage, and prompts a reevaluation of factors affecting specialty choice among medical residents.
(© Jing Miao, Charat Thongprayoon, Oscar Garcia Valencia, Iasmina M Craici, Wisit Cheungpasitporn. Originally published in JMIR Medical Education (https://mededu.jmir.org).)