Using Machine Learning to Assess Physician Competence: A Systematic Review.

Area of Science:

• Medical Education
• Artificial Intelligence
• Health Informatics

Background:

• Physician competence assessment is crucial for patient safety and quality of care.
• Traditional assessment methods may have limitations in objectivity and efficiency.
• Automating assessment using advanced techniques is an emerging area of research.

Purpose of the Study:

• To identify machine learning (ML) techniques applied to physician competence assessment.
• To evaluate how ML techniques assess different competence domains across medical specialties.
• To synthesize current research on ML in medical education and assessment.

Main Methods:

• Systematic literature search across multiple databases (MEDLINE, EMBASE, etc.) up to April 2017.
• Inclusion of studies applying ML techniques to assess physician competence.
• Qualitative narrative synthesis and quality assessment using MERSQI.

Main Results:

• 69 studies met inclusion criteria from 4,953 initial articles.
• General surgery and radiology were the most studied specialties.
• Natural language processing, support vector machines, and hidden Markov models were frequently used ML techniques.
• Patient care and medical knowledge were the most assessed competence domains.

Conclusions:

• Machine learning shows potential for automating physician competence assessment.
• Further validation research is needed to establish the reliability and efficacy of ML techniques.
• ML could enable real-time analysis of physician performance for timely interventions.