Why the study?
The ability of ChatGPT to analyze complex clinical data and provide consistent results was poorly known.
Does ChatGPT-4 provide consistent risk stratification for simulated patients with acute nontraumatic chest pain compared to validated tools?
Does ChatGPT-4 provide consistent risk stratification for simulated patients with acute nontraumatic chest pain compared to validated tools?
ChatGPT-4 shows high mean correlation with established risk scores for atraumatic chest pain but exhibits significant intra-model inconsistency and potential demographic bias, limiting its current clinical reliability.
Answers are grounded in this paper and cite their sources.
Inconsistency and demographic bias limit ChatGPT reliability for chest pain stratification; leaves open need for rigorous AI validation before any clinical use.
BACKGROUND ChatGPT is a large language model with promising healthcare applications. However, its ability to analyze complex clinical data and provide consistent results is poorly known. This study evaluated ChatGPT-4’s risk stratification of simulated patients with acute nontraumatic chest pain compared to validated tools. METHODS Three datasets of simulated case studies were created: one based on the TIMI score variables, another on HEART score variables, and a third comprising 44 randomized variables related to non-traumatic chest pain presentations. ChatGPT independently scored each dataset five times. Its risk scores were compared to calculated TIMI and HEART scores. A model trained on 44 clinical variables was evaluated for consistency. RESULTS ChatGPT showed a high correlation with TIMI and HEART scores (r = 0.898 and 0.928, respectively), but the distribution of individual risk assessments was broad. ChatGPT gave a different risk 45-48% of the time for a fixed TIMI or HEART score. On the 44 variable model, a majority of the five ChatGPT models agreed on a diagnosis category only 56% of the time, and risk scores were poorly correlated (r = 0.605). ChatGPT assigned higher risk scores to males and African Americans. CONCLUSION While ChatGPT correlates closely with established risk stratification tools regarding mean scores, its inconsistency when presented with identical patient data on separate occasions raises concerns about its reliability. The findings suggest that while large language models like ChatGPT hold promise for healthcare applications, further refinement and customization are necessary, particularly in the clinical risk assessment of atraumatic chest pain patients.
No takes yet. Share an insight, caveat, or question.
Heston et al. (2023) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: