Random forests classification analysis for the assessment of diagnostic skill

Area of Science:

• Medical Education
• Rheumatology
• Statistical Analysis

Background:

• Current learning assessment in graduate medical education primarily focuses on individual learners.
• Group-level learning assessment can enhance practice-based learning and support systemic improvements.
• Assessing diagnostic skill alignment is key for effective medical training.

Purpose of the Study:

• To introduce and apply a novel statistical method for assessing learning in graduate medical education.
• To compare the diagnostic skills of rheumatology trainees and attendings using a new analytical approach.
• To evaluate the strength of alignment in diagnostic thinking between trainees and experienced physicians.

Main Methods:

• A random forests classification analysis was employed.
• This method was used to compare the diagnostic skills of rheumatology trainees against rheumatology attendings.
• The analysis captured the strength and correlation of multiple variables to assess thinking alignment.

Main Results:

• The study successfully applied random forests classification to evaluate diagnostic skills.
• The analysis provided insights into the alignment of diagnostic thinking between trainees and attendings.
• The novel statistical approach demonstrated its capability in capturing learning dynamics.

Conclusions:

• Random forests classification offers a novel mechanism for assessing learning and diagnostic skills in graduate medical education.
• This approach can quantify the alignment of thinking between trainees and educators.
• The findings support the use of advanced statistical methods for evaluating learning at both individual and group levels in medical training.