Matt Morgan: Medical assessment in the age of the robot

The ability for doctors to ask the right questions and deal with uncertainty is becoming more important than ever

“Sometimes the questions are complicated and the answers are simple.” — Dr. Seuss

Assessment has long been an important part of the journey to becoming a doctor. The word “finals” is etched into many doctor’s minds even though the old style of a “big bang” exam at the end of training has largely evaporated from most medical training. It has long been recognised that tests of knowledge are important but not a sufficient measure of the process of becoming a safe doctor. However, avoiding tests of pure recall is increasingly important in an age of machine learning and artificial intelligence.

In recent times, the emphasis on tests of pure knowledge recall has transitioned into exam questions focused on knowledge integration. It is not useful to simply know the 10 causes of finger clubbing. The modern student must integrate this knowledge when a patient presents with a constellation of signs and symptoms including finger clubbing. Only then can they choose the correct answer. However, with the development of machines such as IBM’s Watson, even this integration based on knowledge and experience can be simulated and superseded. With the cost of medical training being in excess of £0.5 million per newly graduated doctor, one may question the utility of producing doctors that are less effective at this task than machines.

Of course, the role of a human doctor is much richer than simply providing a diagnosis or a treatment plan. I would argue that my main strength in working as an intensive care consultant is the ability to orchestrate a complex care team to focus on the patient in front of me. This combined with the humanity to communicate with patients, families, and staff is what makes me (currently) more valuable than Watson. However, the quest to become an intensive care consultant is still largely based around tests of physical and cognitive skills that are quickly being superseded by artificial intelligence.

Although undergraduate training has adapted to some of these changes, the postgraduate world needs to embrace the changing roles and skills needed by medical professionals of the next century. No longer will the ability to answer difficult questions be sufficient or even important. Rather, the ability to ask the right questions and deal with uncertainty will be more important than ever. Current medical examinations focus on questions where an answer can provide certainty. However, we need to move instead to a process that assesses a human doctor’s ability to ask and answer questions where there are no right or wrong answers. The ability to ask these difficult questions will be more important than the ability to provide answers. Everyday I encounter issues when people confuse the ability to provide an invasive treatment with the more important question of whether it should be provided at all. Medicine is all about understanding a patient’s story. Thus far robots are not good storytellers but human doctors are.

Thanks to Dr Paul Frost for his feedback on this piece

Matt Morgan is an intensive care consultant, scientist, computer programmer, teacher and geek interested in machine learning, medical education and public engagement. @dr_mattmorgan

Competing interests: I have spoken at a number of education events for which standard travel expenses have been reimbursed. I received a research grant from Heath Research Wales and the Medical Research Council in 2016. I am paid as the lead clinical editor for BMJ’s onExamination.