The Artificial Intelligence Era in Health Economic Modeling
Author(s)
Serena Depalma, MD, MSC1, Chris D. Poole, BSc, PhD2, Paul Turner, PhD2, Rashad Carlton, MSPH, PharmD3.
1Cencora, Amsterdam, Netherlands, 2Cencora, London, United Kingdom, 3Senior Director, Cencora, Conshohocken, PA, USA.
1Cencora, Amsterdam, Netherlands, 2Cencora, London, United Kingdom, 3Senior Director, Cencora, Conshohocken, PA, USA.
OBJECTIVES: Artificial intelligence (AI) technologies are transforming health economic modeling, creating more accurate cost-effectiveness analyses and enabling efficient healthcare resource allocation. This research aims to assess the capabilities of AI in health economics modeling from initial conceptualization to final development and functionality.
METHODS: Two cost-effectiveness Markov models were developed in parallel: one model utilized AI, the other employed traditional development techniques. The models were developed in accordance with the NICE reference case to ensure alignment with established guidelines. This dual approach allowed for a critical comparison of the evaluation metrics of development time, resource use, accuracy, and validation.
RESULTS: AI demonstrated considerable capability in speeding up model conceptualization and design with less resource use but required guidance by a health economist to match health state and transition periods to the traditional model. The model was also developed more efficiently, with AI reducing development time and resource use; integrating data sources to estimate model inputs and designing proxy measures for gaps. Finally, AI created advanced code and formulas for dynamic calculations and suggested user-friendly interfaces. AI platforms support researchers to focus on higher-level strategic tasks, however there were several limitations: namely, requiring an expert guide, validation across various therapeutic and modeling areas and dependence on input format. Several AI platforms were used and outputs produced in Python were consistently the most reliable (compared to Excel and R). Use of free AI tools do not lead to improvements due to restriction on user messaging or AI analysis.
CONCLUSIONS: AI is an innovative resource for future health economics modeling, allowing for faster, more strategic and transparent health economic evaluations. Future research will focus on AI capabilities against development of a HTA model based on a systematic literature review to identify further capabilities in data identification and extraction and address development of interface design.
METHODS: Two cost-effectiveness Markov models were developed in parallel: one model utilized AI, the other employed traditional development techniques. The models were developed in accordance with the NICE reference case to ensure alignment with established guidelines. This dual approach allowed for a critical comparison of the evaluation metrics of development time, resource use, accuracy, and validation.
RESULTS: AI demonstrated considerable capability in speeding up model conceptualization and design with less resource use but required guidance by a health economist to match health state and transition periods to the traditional model. The model was also developed more efficiently, with AI reducing development time and resource use; integrating data sources to estimate model inputs and designing proxy measures for gaps. Finally, AI created advanced code and formulas for dynamic calculations and suggested user-friendly interfaces. AI platforms support researchers to focus on higher-level strategic tasks, however there were several limitations: namely, requiring an expert guide, validation across various therapeutic and modeling areas and dependence on input format. Several AI platforms were used and outputs produced in Python were consistently the most reliable (compared to Excel and R). Use of free AI tools do not lead to improvements due to restriction on user messaging or AI analysis.
CONCLUSIONS: AI is an innovative resource for future health economics modeling, allowing for faster, more strategic and transparent health economic evaluations. Future research will focus on AI capabilities against development of a HTA model based on a systematic literature review to identify further capabilities in data identification and extraction and address development of interface design.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
EE685
Topic
Economic Evaluation
Disease
No Additional Disease & Conditions/Specialized Treatment Areas