Data From: AI Teaches Surgical Diagnostic Reasoning to Medical Students: Evidence from an Experiment Using a Fully Automated, Low-Cost Feedback System

Dataset Info The dataset contains scores across 20 diagnostic video cases, including both trained (intervention group) and untrained conditions. Each row corresponds to a single participant and includes the following fields: Participant ID Group Assignment: Intervention (1: Year-1 students) or Control (2: Year-2 students) Test Time: Immediate post-intervention (1), Delayed post-intervention (2), or Control group (3). All tests were identical. Scores: Binary scores (0: Incorrect or 1: Correct) for identification of each diagnosis in 20 OSVE video cases (e.g., Q1–Q20), where each question represents a unique diagnostic video. Diagnosis Types Covered: Five trained diagnoses (taken by the intervention group)—Cholecystitis, Nephrolithiasis, Gastroenteritis, Pancreatitis, and Appendicitis—and five non-trained diagnoses (not taken by any group) —Gout, Ectopic Pregnancy, Peptic Ulcer, Migraine, and Osteoarthritis. Summative Scores: Total for each individual diagnosis category (e.g., Total-Nephrolithiasis, Total-Cholecystitis, etc.) Total-Others: Sum of scores from 5 non-intervention diagnoses Total-15: The sum of scores from the 15 videos that match the trained diagnoses