Articol
The transformative potential of AI in supporting health economic modeling
-
Derek Swiger, PharmD, MS
-
Paul Turner, PhD, MSc
Explore today’s AI-enabled modeling applications, key limitations, and what’s next for more agile, transparent, and HTA-ready evidence generation.
Economic modeling has long been a cornerstone of healthcare market access, providing the analytical framework to assess cost-effectiveness, budget impact, and value-based pricing strategies. However, as the healthcare landscape grows increasingly complex, the demand for more agile, transparent, and efficient modeling processes has intensified. Artificial intelligence (AI) emerges as a transformative force in this space, streamlining workflows, accelerating decision-making, and enhancing data quality.1
Understanding the current applications of AI in economic modeling, its challenges and limitations, and future opportunities is key to determining how AI is reshaping health economics, and industry can take advantage of these developments while meeting the rigorous expectations of payers and health technology assessment (HTA) agencies.
Current applications of AI in economic modeling
Accelerating model conceptualization and design
AI is revolutionizing the initial stages of economic modeling by enabling rapid prototyping of model structures.2 Traditionally, health economists were often constrained by labor-intensive processes that locked them into a single model design. AI tools now allow teams to test multiple model structures simultaneously, identifying the most efficient and credible options to support a product’s value story.
“Having tools that can allow us to very rapidly prototype models enables us to understand the optimal model design for products to identify and support the value story,” said Chris Poole, PhD, a senior health economist at Cencora.
This ability to test diverse model designs early in the process supports the goal of finding the right balance of uncertainty and complexity and ensures that the chosen structure aligns with payer expectations. This is not only time-saving but also a huge cost-savings since the delays in submissions and the risk that agencies will reject complex, poorly justified models can cost companies millions.
“Having tools that can allow us to very rapidly prototype models enables us to understand the optimal model design for products to identify and support the value story,” said Chris Poole, PhD, a senior health economist at Cencora.
This ability to test diverse model designs early in the process supports the goal of finding the right balance of uncertainty and complexity and ensures that the chosen structure aligns with payer expectations. This is not only time-saving but also a huge cost-savings since the delays in submissions and the risk that agencies will reject complex, poorly justified models can cost companies millions.
Filling data gaps and enhancing data quality
Data gaps are a persistent challenge in economic modeling, particularly for rare diseases or novel therapies where there are no data, not enough data, or where the data available are not of good enough quality. AI is proving invaluable in consolidating disparate data sources and simulating missing inputs. For example, AI can digitize Kaplan-Meier (KM) curves and generate individual participant data for survival analyses, enabling more robust parameterization of survival data.3
“It’s astonishing that you can take a published KM curve and show that to a genAI tool with access to a Python container to execute the code in, and it will digitize the curve, write the code, and execute it so that it can the simulate individual participant data (IPD) populations that would have given you that curve,” Dr. Poole said, drawing on his experience with AI tools. “It’s extraordinary. The best part is that it’s easy to validate, you just work backwards by using the IPD to generate the KM curve.”
This capability dramatically reduces the time required for statistical analysis, allowing health economists to focus on higher-value tasks. However, ensuring transparency and validity in AI-driven data simulations remains critical.
“It’s astonishing that you can take a published KM curve and show that to a genAI tool with access to a Python container to execute the code in, and it will digitize the curve, write the code, and execute it so that it can the simulate individual participant data (IPD) populations that would have given you that curve,” Dr. Poole said, drawing on his experience with AI tools. “It’s extraordinary. The best part is that it’s easy to validate, you just work backwards by using the IPD to generate the KM curve.”
This capability dramatically reduces the time required for statistical analysis, allowing health economists to focus on higher-value tasks. However, ensuring transparency and validity in AI-driven data simulations remains critical.
Streamlining report authoring
AI is also transforming how technical reports and other documents, such as budget impact model reports for Academy of Managed Care Pharmacy dossiers, are authored.4 By automating repetitive tasks like formatting and data integration, AI allows health economists to allocate more time to ensuring a model best supports the value story of the product.
The labor-saving potential of AI is noteworthy. Within Cencora, for example, we are building internal AI systems to quickly draft numerous types of reports with the goal of applying these systems to our own economic modeling reports in the near future. It’s not just about creating new reports. Content reuse between different HTA agencies is also ripe for automation.
“A lot of these templates from NICE (National Institute for Health and Care Excellence) or other agencies might be structured or ordered differently, but the information needed to complete them is identical,” said Derek Swiger, PharmD, MS, a digital innovation expert in market access and health economics and outcomes research (HEOR). “We can certainly borrow a page from what folks are doing with AI in the clinical and regulatory spaces with respect to content modularization and reuse. An example is how the drafting of clinical study reports is being automated by AI tools through reuse of content from the underlying documents like the study protocol.”
This shift not only enhances efficiency but also ensures consistency across documents tailored to diverse payer and regulatory requirements.
The labor-saving potential of AI is noteworthy. Within Cencora, for example, we are building internal AI systems to quickly draft numerous types of reports with the goal of applying these systems to our own economic modeling reports in the near future. It’s not just about creating new reports. Content reuse between different HTA agencies is also ripe for automation.
“A lot of these templates from NICE (National Institute for Health and Care Excellence) or other agencies might be structured or ordered differently, but the information needed to complete them is identical,” said Derek Swiger, PharmD, MS, a digital innovation expert in market access and health economics and outcomes research (HEOR). “We can certainly borrow a page from what folks are doing with AI in the clinical and regulatory spaces with respect to content modularization and reuse. An example is how the drafting of clinical study reports is being automated by AI tools through reuse of content from the underlying documents like the study protocol.”
This shift not only enhances efficiency but also ensures consistency across documents tailored to diverse payer and regulatory requirements.
Challenges and limitations of AI in economic modeling
AI’s potential in health economic modeling is extensive, but challenges and limitations remain. There are often discussions in the media about AI replacing people, but it is important to note that while AI excels at generating options and automating processes, it cannot replace human expertise in strategic decision-making. Health economists play a critical role in ensuring that AI-generated models are feasible, credible, and aligned with clinical and reimbursement realities.
Paul Turner, PhD, a senior health economist at Cencora, emphasized the importance of human involvement:
“AI can generate the options for you, but there does need to be an element of human involvement to make sure what’s picked is feasible. That oversight is critical.”
This human-in-the-loop approach is essential to maintaining the integrity of economic models and ensuring their suitability for decision-making. HEOR teams must also balance the use of AI to create highly complex models with the risks. While intricate models may meet payer expectations, they often introduce additional parameter uncertainty and raise potential issues with agencies.
“Just because we can use AI to build complex models doesn’t mean we should,” noted Dr. Poole. “For example, the HTA agency might not find the model suitable for decision-making.”
Striking the right balance between complexity and uncertainty is key to maximizing the utility of AI-driven models.
Paul Turner, PhD, a senior health economist at Cencora, emphasized the importance of human involvement:
“AI can generate the options for you, but there does need to be an element of human involvement to make sure what’s picked is feasible. That oversight is critical.”
This human-in-the-loop approach is essential to maintaining the integrity of economic models and ensuring their suitability for decision-making. HEOR teams must also balance the use of AI to create highly complex models with the risks. While intricate models may meet payer expectations, they often introduce additional parameter uncertainty and raise potential issues with agencies.
“Just because we can use AI to build complex models doesn’t mean we should,” noted Dr. Poole. “For example, the HTA agency might not find the model suitable for decision-making.”
Striking the right balance between complexity and uncertainty is key to maximizing the utility of AI-driven models.
Future trends and opportunities
The potential for AI in health economics continues to grow with the advent of newer systems and capabilities. For example, agentic AI systems are ones that are capable of autonomously updating model parameters with the latest evidence, accomplished through linking with a “living” systematic literature review system. These “living” models can provide real-time insights, improve forecast accuracy, and enable more agile decision-making. We believe that agentic AI systems are poised to redefine economic modeling.
Another area where we expect AI systems to become instrumental is by analyzing historical HTA submissions and payer feedback to identify risk factors associated with model rejection or inefficiencies.
Dr. Poole made the following prediction:
“For any given disease area, I think we will probably see AI being used to create a database of factors that lead to success in a model being regarded as suitable for decision-making and, crucially, what factors are associated with a model being deemed not suitable. Maybe there are 5 PhDs in that alone with traditional methods, but an AI system could probably support a really good tool to identify those risk factors."
We predict these tools will help manufacturers proactively address potential pitfalls in their submissions, improving the likelihood of successful outcomes.
Another area where we expect AI systems to become instrumental is by analyzing historical HTA submissions and payer feedback to identify risk factors associated with model rejection or inefficiencies.
Dr. Poole made the following prediction:
“For any given disease area, I think we will probably see AI being used to create a database of factors that lead to success in a model being regarded as suitable for decision-making and, crucially, what factors are associated with a model being deemed not suitable. Maybe there are 5 PhDs in that alone with traditional methods, but an AI system could probably support a really good tool to identify those risk factors."
We predict these tools will help manufacturers proactively address potential pitfalls in their submissions, improving the likelihood of successful outcomes.
Conclusion
AI is reshaping the landscape of economic modeling, offering unprecedented speed, efficiency, and accuracy. From accelerating model conceptualization to filling data gaps and automating report authoring, AI is transforming how health economists approach market access challenges.
However, the adoption of AI is not without its challenges. Human oversight, transparency, and strategic decision-making remain critical to ensuring the validity and feasibility of AI-driven models. As we look to the future, the integration of agentic AI systems will unlock new opportunities for evidence generation.
However, the adoption of AI is not without its challenges. Human oversight, transparency, and strategic decision-making remain critical to ensuring the validity and feasibility of AI-driven models. As we look to the future, the integration of agentic AI systems will unlock new opportunities for evidence generation.
*Sursele continuă mai jos
Precizare:
Informațiile furnizate în acest articol nu constituie consultanță juridică. Cencora, Inc. încurajează insistent cititorii să revizuiască informațiile disponibile legate de subiectele discutate și să se bazeze pe propria experiență și expertiză în luarea deciziilor legate de acestea.
Luați legătura cu echipa noastră
Echipa noastră de experți în valoare este dedicată transformării dovezilor, informațiilor privind politicile și informațiilor de piață în strategii eficiente de acces la piața globală. Permiteți-ne să vă ajutăm să navigați cu încredere prin peisajul complex al asistenței medicale din ziua de azi. Contactați-ne pentru a afla în ce mod vă putem sprijini în atingerea obiectivelor.
Sources
1. Fleurence RL, Bian J, Wang X, et al; ISPOR Working Group on Generative AI. Generative artificial intelligence for health technology assessment: opportunities, challenges, and policy considerations: an ISPOR working group report. Value Health. 2025;28(2):175-183. doi:10.1016/j.jval.2024.10.3846
2. Depalma S, Poole CD, Turner P, Carlton R. The artificial intelligence era in health economic modeling. Poster presented at: ISPOR Europe 2025; November 9-12. Glasgow, Scotland. EE685. https://www.ispor.org/heor-resources/presentations-database/presentation-cti/ispor-europe-2025/poster-session-5-2/the-artificial-intelligence-era-in-health-economic-modeling
3. Annan A, Mojarad MR, Du J, Xu Y. Automated extraction of Kaplan-Meier survival curves using generative artificial intelligence and computer vision. Presented at: ISPOR 2025; May 13-16; Montréal, Quebec, CA. MSR33. https://www.valueinhealthjournal.com/article/S1098-3015(25)01310-5/fulltext
4. Fleurence RL, Wang X, Bian J, et al; ISPOR Working Group on Generative AI. A taxonomy of generative artificial intelligence in health economics and outcomes research: an ISPOR working group report. Value Health. 2025;28(11):1601-1610. doi:10.1016/j.jval.2025.04.2167
