Mentee
Radhika Gupta, BS
Perelman School of Medicine
Mentor
Albert Yan, MD
Children’s Hospital of Philadelphia
Overview
Alopecia areata (AA) is a condition in which the immune system attacks hair follicles resulting in hair loss. To track progression of hair loss pre- and post-treatment, physicians use the Severity of Alopecia Tool (SALT) which quantifies hair loss in all 4 scalp quadrants (right, left, top, back); however, physician-related subjectivity exists in SALT scoring resulting in interobserver variability. Here, we aim to assess ChatGPT’s ability to generate SALT scores of patients’ scalps based on images. Several databases will be queried to evaluate de-identified 4-view images of patients’ scalps with corresponding clinician-derived SALT scores. After verifying ChatGPT’s understanding of SALT, 4-view images from 150 patients’ scalps will be uploaded and ChatGPT will be prompted to generate SALT scores for each patient. To determine how well ChatGPT performs in comparison to the clinician-derived SALT scores, Microsoft Excel’s ANOVA function will be used to calculate an intraclass correlation coefficient which is a measure of concordance.
Status
This project was funded through a 2024 PeDRA Emerging Investigator Research Grant.
