Grant Number:	1R01CA300393-01A1 Interpret this number
Primary Investigator:	Rutter, Carolyn
Organization:	Fred Hutchinson Cancer Center
Project Title:	Methods for Microsimulation Model Calibration
Fiscal Year:	2026

PROJECT SUMMARY Microsimulation modeling is a powerful tool for using scientific evidence to inform health policy by projecting the long-term harms and benefits of interventions. Models are especially useful for estimating long-term effectiveness, harms, and costs of new cancer screening tests because gold-standard studies take decades to complete. Out of necessity, modeling has become an accepted way to generate evidence to aid in the creation of screening guidelines set by United States Preventive Services Task Force. The stakes are high: by informing policy, modeling affects the care offered to millions of Americans. Given the importance of simulation models, the lack of guidance on their development is both surprising and concerning. There are large gaps in our understanding of how to select model parameters so that model projections are consistent with statistics that describe the disease process, including results from clinical trials, findings from observational studies, and incidence and mortality rates from registries. This process, called calibration, is an essential step in developing a microsimulation model. Our proposed research directly addresses this knowledge gap by providing a flexible, publicly available tool for model calibration (Aim 1), developing a framework and tools for evaluating calibration methods (Aim 2), and developing and distributing guidelines for model calibration (Aim 3). This work is highly responsive to the notice of funding opportunity PA-25-172, which has a goal of supporting cancer research in statistical and analytic methods including “decision modeling using simulation or other methods to determine efficient or cost-effective strategies for the prevention, early detection, or treatment of cancer.” The tools we will create and distribute for evaluating calibration algorithms will be general and useful for assessing and comparing a wide range of calibration algorithms. Our proposed work will catalyze development of calibration methods that are sorely needed by the modeling community.