Professional vision and medical education: dividing diagnoses into categories to support more effective learning in dermatology

Background Proficiency in interpretation of visual findings, professional vision, plays an essential role in the doctor's profession. Thus, learning visual features from images is a key curricular component of medical education, for example, in radiology and dermatology. Nevertheless, there is scarcity of research on professional vision in dermatology. Aims The aims of this study were to 1) measure how professional vision differs before and after a dermatology course, and if there are differences across diagnoses, 2) cluster different diagnoses based on learning patterns, 3) investigate the development of students’ accuracy, decision times and confidence within different categories. Sample Participants were 394 medical students and 7 experts. Methods Participants completed an online perceptual learning module at three timepoints (pre, mid and post) during a course. The task was to recognize a visually presented dermatological condition and choose the correct diagnosis from eight options and rate the confidence in decision. The module contained 47 different diagnoses, each presented once. Students repeated the modules multiple times. A two-parameter Item-Response-Theory model was fitted to the data. Based on item characteristic curves, the diagnoses were clustered into eight categories. Results Diagnostic accuracy, decision times and confidence improved in most, but not all, pre-mid and mid-post comparisons. There were major differences in learning patterns (amount and pace of improvement) across the eight categories. Conclusions The differences between diagnosis categories were larger than differences between individual participants. This finding can be utilized in medical education by focusing teaching resources on diagnoses that are difficult to learn.