Diagnosing virtual patients in a technology-rich learning environment: a sequential Mining of Students’ efficiency and behavioral patterns

This study examined the relationships between clinical reasoning behaviors and diagnostic efficiency in the context of diagnosing a virtual patient in BioWorld, a technology-rich environment designed for medical students to practice clinical reasoning skills. Eighty-two medical students who correctly solved a patient case with Diabetes Mellitus were included in this study. These students were grouped into efficient and less efficient groups based on the time they spent diagnosing the case using k-means clustering. Students’ clinical reasoning behaviors were recorded in log files and further coded as either relevant or irrelevant to the final correct diagnosis. Independent t-tests and sequential pattern mining were then conducted to compare the differences between efficient and less efficient groups. Results revealed that students in the less efficient group collected significantly more irrelevant evidence, ordered more lab tests, and proposed more incorrect hypotheses than efficient students. Moreover, less efficient students demonstrated more disorganized behavioral patterns than efficient students. These findings underscored metacognitive skills in delivering an efficient diagnosis. This study also informs the practice of medical education in terms of the development of expertise, as well as the design of interventions and scaffolding in promoting efficient learning or clinical reasoning.