ChatGPT and post-test probability

TL;DR

This paper evaluates ChatGPT's ability to perform probabilistic medical diagnostic reasoning, specifically Bayesian updating, and explores how prompt engineering can improve its accuracy in this domain.

Contribution

It systematically probes ChatGPT's performance in Bayesian medical diagnosis tasks and demonstrates how prompt engineering can mitigate errors.

Findings

ChatGPT makes errors when using medical variable names.

Prompt engineering can partially reduce these errors.

Results inform future research on LLMs in healthcare.

Abstract

Reinforcement learning-based large language models, such as ChatGPT, are believed to have potential to aid human experts in many domains, including healthcare. There is, however, little work on ChatGPT's ability to perform a key task in healthcare: formal, probabilistic medical diagnostic reasoning. This type of reasoning is used, for example, to update a pre-test probability to a post-test probability. In this work, we probe ChatGPT's ability to perform this task. In particular, we ask ChatGPT to give examples of how to use Bayes rule for medical diagnosis. Our prompts range from queries that use terminology from pure probability (e.g., requests for a posterior of A given B and C) to queries that use terminology from medical diagnosis (e.g., requests for a posterior probability of Covid given a test result and cough). We show how the introduction of medical variable names leads to an increase in the number of errors that ChatGPT makes. Given our results, we also show how one can use prompt engineering to facilitate ChatGPT's partial avoidance of these errors. We discuss our results in light of recent commentaries on sensitivity and specificity. We also discuss how our results might inform new research directions for large language models.