Purpose Only from reviewing applications, it is difficult to identify which applicants will be successful ophthalmology residents. The change of the USMLE Step 1 scoring to "Pass/Fail" removes another quantitative metric. We aimed to identify application attributes correlated with successful residency performance. This study also used artificial intelligence (AI) to evaluate letters of recommendation (LOR), the Dean's letter (MSPE), and personal statement (PS). Design Retrospective analysis of application characteristics versus residency performance was conducted. Participants Residents who graduated from the Dean McGee Eye Institute/University of Oklahoma Ophthalmology residency from 2004 to 2019 were included in this study. Methods Thirty-four attributes were recorded from each application. Residents were subjectively ranked into tertiles and top and bottom deciles based on residency performance by faculty present during their training. The Ophthalmic Knowledge Assessment Program (OKAP) examination scores were used as an objective performance metric. Analysis was performed to identify associations between application attributes and tertile/decile ranking. Additional analysis used AI and natural language processing to evaluate applicant LORs, MSPE, and PS. Main Outcome Measures Characteristics from residency applications that correlate with resident performance were the primary outcome of this study. Results Fifty-five residents and 21 faculty members were included. A grade of "A" or "Honors" in the obstetrics/gynecology (OB/GYN) clerkship and the presence of a home ophthalmology department were associated with ranking in the top tertile but not the top decile. Mean core clerkship grades, medical school ranking in the top 25 U.S. News and World Report (USNWR) primary care rankings, and postgraduate year (PGY)-2 and PGY-3 OKAP scores were predictive of being ranked in both the top tertile and the top decile. USMLE scores, alpha-omega-alpha (AOA) status, and number of publications did not correlate with subjective resident performance. AI analysis of LORs, MSPE, and PS did not identify any text features that correlated with resident performance. Conclusions Many metrics traditionally felt to be predictive of residency success (USMLE scores, AOA status, and research) did not predict resident success in our study. We did confirm the importance of core clerkship grades and medical school ranking. Objective measures of success such as PGY-2 and PGY-3 OKAP scores were associated with high subjective ranking.
Keywords: Residency Application; artificial intelligence; resident performance.
The Author(s). This is an open access article published by Thieme under the terms of the Creative Commons Attribution-NonDerivative-NonCommercial License, permitting copying and reproduction so long as the original work is given appropriate credit. Contents may not be used for commercial purposes, or adapted, remixed, transformed or built upon. ( https://creativecommons.org/licenses/by-nc-nd/4.0/ ).