Back

Diagnostic Performance of ChatGPT to perform emergency department triage: A systematic review and meta-analysis

Kaboudi, N.; Firouzbakht, S.; Shahir Eftekhar, M.; Fayazbakhsh, F.; Joharivarnoosfaderani, N.; Ghaderi, S.; Dehdashti, M.; Mohtasham Kia, Y.; Afshari, M.; Vasaghi-Gharamaleki, M.; Haghani, L.; Khalaj, F.; Mohammadi, Z.; Hasanabadi, Z.; Shahidi, R.

2024-05-20 emergency medicine
10.1101/2024.05.20.24307543 medRxiv
Show abstract

BackgroundArtificial intelligence (AI), particularly ChatGPT developed by OpenAI, has shown potential in improving diagnostic accuracy and efficiency in emergency department (ED) triage. This study aims to evaluate the diagnostic performance and safety of ChatGPT in prioritizing patients based on urgency in ED settings. MethodsA systematic review and meta-analysis were conducted following PRISMA guidelines. Comprehensive literature searches were performed in Scopus, Web of Science, PubMed, and Embase. Studies evaluating ChatGPTs diagnostic performance in ED triage were included. Quality assessment was conducted using the QUADAS-2 tool. Pooled accuracy estimates were calculated using a random-effects model, and heterogeneity was assessed with the I{superscript 2} statistic. ResultsFourteen studies with a total of 1,412 patients or scenarios were included. ChatGPT 4.0 demonstrated a pooled accuracy of 0.86 (95% CI: 0.64-0.98) with substantial heterogeneity (I{superscript 2} = 93%). ChatGPT 3.5 showed a pooled accuracy of 0.63 (95% CI: 0.43-0.81) with significant heterogeneity (I{superscript 2} = 84%). Funnel plots indicated potential publication bias, particularly for ChatGPT 3.5. Quality assessments revealed varying levels of risk of bias and applicability concerns. ConclusionChatGPT, especially version 4.0, shows promise in improving ED triage accuracy. However, significant variability and potential biases highlight the need for further evaluation and enhancement.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.