In Reply to Nagirimadugu and Tippireddy

Abstract

We are pleased to read the comments from Nagirimadugu and Tippireddy because the perspective of medical students is imperative to successful integration of machine learning (ML) content into medical curricula. We agree that a multimodal approach to teaching ML is integral to effective incorporation of ML content into curricula. In addition to lectures and small groups, learners would benefit from synchronous and asynchronous learning using technology to provide foundational ML content and reinforce concepts. ML content must also be incorporated across the medical education continuum. Education of house officers and faculty physicians is necessary to enhance the likelihood that important conversations related to the strengths and limitations of ML take place in clinical settings. Evidence-based medicine, which emphasizes incorporation of best research evidence, patient values and preferences, and clinician expertise, provides a proven framework for critically evaluating literature for risk of bias and applicability. Progress in the development of frameworks for critically evaluating studies and applications that include ML algorithms is only beginning. If ML is to be used in practice with regularity, users’ guides must be developed to allow clinicians to critically evaluate ML algorithms and teach learners to do the same. 1 Although the risk of racial and social biases is now recognized, development of best practice guides will require an interprofessional effort that includes expertise from clinicians, computer scientists, engineers, statisticians, and epidemiologists. And let us not forget patients. As noted by Nagirimadugu and Tippireddy, incorporation of ML into curricula provides an opportunity for interprofessional relationships between these stakeholders. Like how the nonclinical anatomist or biochemist currently plays significant roles in medical education, computer scientists and engineers will have equal—if not more important—roles in the medical education of the future. These relationships should extend beyond the formal teacher–learner relationships and into learner–learner alliances. Medical students and computer science students certainly have a lot to learn from one another in this domain.

Ähnliche Arbeiten