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

Area of Science:

• Epidemiology
• Biostatistics
• Geographic Information Systems

Background:

• Cancer registries often suppress small-area counts to protect patient privacy.
• Accurate small-area cancer data is crucial for public health surveillance and resource allocation.
• Existing methods may lack precision or fail to account for demographic variations.

Purpose of the Study:

• To develop and validate an innovative multi-constraint Monte Carlo simulation method.
• To estimate suppressed county-level cancer counts for population subgroups.
• To extend cancer data downscaling from county to ZIP Code Tabulation Areas (ZCTA).

Main Methods:

• Employed a multi-constraint Monte Carlo simulation using population structure as probability.
• Utilized known cancer counts at higher geographic levels and demographic groups as constraints.
• Validated the approach using 2016-2020 cancer incidence data from the Utah Cancer Registry.

Main Results:

• The method accurately estimates suppressed cancer counts with high precision.
• Results demonstrate consistency across urban-rural populations.
• Outperformed Random Forest and Extreme Gradient Boosting machine learning models.

Conclusions:

• The Monte Carlo simulation method offers a robust approach for small-area cancer data estimation.
• This technique enhances cancer surveillance by providing reliable ZCTA-level estimates.
• The method effectively balances data privacy with the need for granular public health information.