Evaluating the accuracy of a state-of-the-art large language model for prediction of admissions from the emergency room

• Published in: Journal of the American Medical Informatics Association : JAMIA 2024-09, Vol.31 (9), p.1921-1928
• Main Authors: Glicksberg, Benjamin S, Timsina, Prem, Patel, Dhaval, Sawant, Ashwin, Vaid, Akhil, Raut, Ganesh, Charney, Alexander W, Apakama, Donald, Carr, Brendan G, Freeman, Robert, Nadkarni, Girish N, Klang, Eyal
• Format: Article
• Language: English
• Subjects: Artificial Intelligence; Electronic Health Records; Emergency Service, Hospital; Humans; Machine Learning; Natural Language Processing; Patient Admission; Retrospective Studies; Supervised Machine Learning
• ISSN: 1067-5027; 1527-974X; 1527-974X
• DOI: 10.1093/jamia/ocae103

Artificial intelligence (AI) and large language models (LLMs) can play a critical role in emergency room operations by augmenting decision-making about patient admission. However, there are no studies for LLMs using real-world data and scenarios, in comparison to and being informed by traditional supervised machine learning (ML) models. We evaluated the performance of GPT-4 for predicting patient admissions from emergency department (ED) visits. We compared performance to traditional ML models both naively and when informed by few-shot examples and/or numerical probabilities. We conducted a retrospective study using electronic health records across 7 NYC hospitals. We trained Bio-Clinical-BERT and XGBoost (XGB) models on unstructured and structured data, respectively, and created an ensemble model reflecting ML performance. We then assessed GPT-4 capabilities in many scenarios: through Zero-shot, Few-shot with and without retrieval-augmented generation (RAG), and with and without ML numerical probabilities. The Ensemble ML model achieved an area under the receiver operating characteristic curve (AUC) of 0.88, an area under the precision-recall curve (AUPRC) of 0.72 and an accuracy of 82.9%. The naïve GPT-4's performance (0.79 AUC, 0.48 AUPRC, and 77.5% accuracy) showed substantial improvement when given limited, relevant data to learn from (ie, RAG) and underlying ML probabilities (0.87 AUC, 0.71 AUPRC, and 83.1% accuracy). Interestingly, RAG alone boosted performance to near peak levels (0.82 AUC, 0.56 AUPRC, and 81.3% accuracy). The naïve LLM had limited performance but showed significant improvement in predicting ED admissions when supplemented with real-world examples to learn from, particularly through RAG, and/or numerical probabilities from traditional ML models. Its peak performance, although slightly lower than the pure ML model, is noteworthy given its potential for providing reasoning behind predictions. Further refinement of LLMs with real-world data is necessary for successful integration as decision-support tools in care settings.