Random Forests Classification Analysis for the Assessment of Diagnostic Skill

Mechanisms are needed to assess learning in the context of graduate medical education. In general, research in this regard is focused on the individual learner. At the level of the group, learning assessment can also inform practice-based learning and may provide the foundation for whole systems improvement. The authors present the results of a random forests classification analysis of the diagnostic skill of rheumatology trainees as compared with rheumatology attendings. A random forests classification analysis is a novel statistical approach that captures the strength of alignment of thinking between student and teacher. It accomplishes this by providing information about the strength and correlation of multiple variables.