Algorithms in clinical decision making

This is a hybrid event.

Evidence-based algorithms are developed in ever-increasing numbers to support decisions in several domains including healthcare. Nevertheless, the literature on “algorithm aversion”, i.e., reluctance to use superior but imperfect algorithms, and “egocentric advice discounting”, i.e., overweighting our own estimates and underweighting advice, cast doubt on their promise of improving decision making. I will present a recent programme of work, funded by Cancer Research UK, that investigated whether an unnamed cancer risk algorithm, which calculates the probability that a patient with symptoms has a cancer, can influence GPs’ risk estimates and referral decisions. In a series of online vignette-based experiments, we also tested interventions to promote the algorithm’s influence on clinical judgement. Across studies, we found little evidence of egocentric advice discounting. Instead, our findings paint a more optimistic picture about the potential of risk algorithms to influence medical decisions and improve clinical risk assessment.

Olga Kostopoulou studied psychology at the National and Kapodistrian University of Athens, Greece. With funding from the State Scholarships Foundation of Greece, she continued her studies in the UK, obtaining an MSc in Applied Psychology and a PhD in Psychology from Cardiff University. She has done extensive research on medical diagnosis and diagnostic error from a cognitive psychology perspective, and investigates how to support clinical decisions. Olga leads the Decision Psychology research group at Imperial College London, and she was the 2021-22 President of the Society for Medical Decision Making.


Meeting ID: 827 5284 2985
Passcode: 052890