OBJECTIVES: Artificial Intelligence (AI) and Machine Learning (ML) are revolutionizing Health Economics and Outcomes Research (HEOR) by providing powerful tools for screening, extracting, and summarizing literature, analyzing large datasets, improving predictive accuracy, and optimizing healthcare delivery. This review aims to summarize the current use of AI/ML applications across different HEOR domains, emphasizing their impact on healthcare decision-making.

METHODS: A targeted review of studies that utilized AI/ML techniques in HEOR was conducted in Medline from 2021 onwards. Additionally, Value in Health journal was manually searched using keywords for AI, ML, natural language processing and large language model (LLM). Identified studies were categorized by their use across HEOR domains.

RESULTS: 105 studies were identified (16 full-texts and 89 abstracts), of which AI/ML applications were primarily used in systematic literature reviews (SLRs) (n=60), followed by real-world evidence (RWE) (n=25), health technology assessment (HTA) methods (n=13), and economic evaluations (n=7). In SLRs, 80% of studies focused on applications aimed at enhancing screening efficiency and data extraction. In RWE, LLMs were predominantly used to analyze electronic health records and claims data, predict outcomes (such as comparative effectiveness, medication adherence, quality of life), automate protocol writing, and improve coding efficiency. AI applications in economic evaluations focused on model automation, input validation, plot digitization, code writing, and model conceptualization. Additionally, some studies explored simulating HTA committee discussions and predicting HTA outcomes using AI/ML techniques.

CONCLUSIONS: AI/ML integration in HEOR shows substantial potential to enhance healthcare decision-making and efficiency. The predominant use in SLRs highlights their capacity to streamline research processes. However, no guidelines currently exist for using AI/ML techniques in HEOR or their acceptance by HTA agencies. Developing guidelines is recommended to standardize AI/ML applications, thereby improving decision-making processes. Future research should address these challenges and continue exploring AI/ML applications in HEOR to maximize their benefits.