A multi-constraint Monte Carlo Simulation approach to downscaling cancer data

Abstract

This study employs an innovative multi-constraint Monte Carlo simulation method to estimate suppressed county-level cancer counts for population subgroups and extend the downscaling from county to ZIP Code Tabulation Areas (ZCTA) in the U.S. Given the known cancer counts at a higher geographic level and larger demographic groups at the same geographic level as constraints, this method uses the population structure as probability in the Monte Carlo simulation process to estimate suppressed data entries. It not only ensures consistency across various data levels but also accounts for demographic structure that drives varying cancer risks. The 2016–2020 cancer incidence data from the Utah Cancer Registry is used to validate our approach. The method yields results with high precision and consistency across the full urban-rural continuum, and significantly outperforms several machine-learning models such as Random Forest and Extreme Gradient Boosting.