Is deep learning in medical education possible?

Source: KevinMD

Despite recent buzz about shifting resident education to community health centers, hospital based education is here to stay. The model of education, though outmoded, is simple. Get residents exposed to as much disease as possible, in the shortest amount of time. The future of American health care is not in acute management of tertiary care; but in integrated, team-based care. To get there involves focusing not only on educational content, but also on the process of how we teach and learn.


Consider a typical inpatient medicine service, with ten members to a team: an attending, senior resident, three interns, three medical students, a pharmacist and a pharmacy student.

First, some numbers:

Combined years of training: 50
Hours spent on clinical rounds (weekly): 30
Hours devoted to education (weekly): 5
Hours devoted to team building, quality improvement, and longitudinal care: 0

The numbers speak for themselves. Clearly, as a system, we have chosen to devalue team-based education in favor of teaching isolated knowledge to different training levels. This is the issue of content. But what about the process of rounding? Two simple concepts would go a long way.

1. Teach clinicians cognitive skills. This idea comes from the Right Question Institute, an organization dedicated to teaching students to ask critical questions. The cognitive skills involve three types of thinking:

Convergent thinking.  This involves integrating a host of information for a single purpose. For instance, looking at a list of medication and deciding which contributes to delirium. 

Divergent thinking. This looks at the process from the opposing spectrum: given that a patient has confusion, what medications could be causing this? 

Metacognition.  This is the process by which we “think about thinking” — evaluating, judging, and gauging our strategies to approach clinical thinking. This could involve asking: we approached the diagnosis of heart failure physiologically; how else could we have approached it?”

Hospital-based models of care involves all three capacities, though highly skews towards convergent thinking. The divergent process of generating differential diagnoses is quickly losing importance as technological tests proliferate. If you can read the CT scan of a patient with abdominal pain prior to seeing them in the flesh, you need not think divergently. The highest yield change may be to teach tools of metacognition, as these lead to new approaches to patient care, and sparks curiosity.

2. Create opportunities for deep learning. One model of education describes learners as superficial, strategic and deep. The superficial learner does purely what is needed to get by — in the case of the hospital intern, maximizing the efficiency of computerized orders. The strategic learner focuses on what he needs to get ahead; presenting information clearly and looking up facts about a case. The deep learner, in contrast, often asks bigger “why” questions about approach to medication management, physiology, or the social lives of patients. Incentivizing deep learning involves permitting time for intensive reading, encouraging development of multiple cognitive skills, and mobilizing knowledge from each team member.

As healthcare transitions to a team-based model of care, medical education must follow. Creating conditions for deep learning may be this leverage point, if we dare to think differently.

Tom Peteet is an internal medicine resident.