RESUMO
BACKGROUND: Current generation large language models (LLMs) such as Generative Pre-Trained Transformer 4 (GPT-4) have achieved human-level performance on many tasks including the generation of computer code based on textual input. This study aimed to assess whether GPT-4 could be used to automatically programme two published health economic analyses. METHODS: The two analyses were partitioned survival models evaluating interventions in non-small cell lung cancer (NSCLC) and renal cell carcinoma (RCC). We developed prompts which instructed GPT-4 to programme the NSCLC and RCC models in R, and which provided descriptions of each model's methods, assumptions and parameter values. The results of the generated scripts were compared to the published values from the original, human-programmed models. The models were replicated 15 times to capture variability in GPT-4's output. RESULTS: GPT-4 fully replicated the NSCLC model with high accuracy: 100% (15/15) of the artificial intelligence (AI)-generated NSCLC models were error-free or contained a single minor error, and 93% (14/15) were completely error-free. GPT-4 closely replicated the RCC model, although human intervention was required to simplify an element of the model design (one of the model's fifteen input calculations) because it used too many sequential steps to be implemented in a single prompt. With this simplification, 87% (13/15) of the AI-generated RCC models were error-free or contained a single minor error, and 60% (9/15) were completely error-free. Error-free model scripts replicated the published incremental cost-effectiveness ratios to within 1%. CONCLUSION: This study provides a promising indication that GPT-4 can have practical applications in the automation of health economic model construction. Potential benefits include accelerated model development timelines and reduced costs of development. Further research is necessary to explore the generalisability of LLM-based automation across a larger sample of models.
RESUMO
BACKGROUND: The emergence of artificial intelligence, capable of human-level performance on some tasks, presents an opportunity to revolutionise development of systematic reviews and network meta-analyses (NMAs). In this pilot study, we aim to assess use of a large-language model (LLM, Generative Pre-trained Transformer 4 [GPT-4]) to automatically extract data from publications, write an R script to conduct an NMA and interpret the results. METHODS: We considered four case studies involving binary and time-to-event outcomes in two disease areas, for which an NMA had previously been conducted manually. For each case study, a Python script was developed that communicated with the LLM via application programming interface (API) calls. The LLM was prompted to extract relevant data from publications, to create an R script to be used to run the NMA and then to produce a small report describing the analysis. RESULTS: The LLM had a > 99% success rate of accurately extracting data across 20 runs for each case study and could generate R scripts that could be run end-to-end without human input. It also produced good quality reports describing the disease area, analysis conducted, results obtained and a correct interpretation of the results. CONCLUSIONS: This study provides a promising indication of the feasibility of using current generation LLMs to automate data extraction, code generation and NMA result interpretation, which could result in significant time savings and reduce human error. This is provided that routine technical checks are performed, as recommend for human-conducted analyses. Whilst not currently 100% consistent, LLMs are likely to improve with time.
