SHAPE: Spatial Health and Population Estimator

A Spatial Microsimulation Framework for Public Health Synthetic Population Generation and Small-Area Estimation (SAE)

Programming Language: R

SHAPE GIS Web App: Interactively explore SHAPE’s estimates, and visualize health outcomes and behaviors across space and time - https://tinyurl.com/SHAPE-GIS

OSF Data Repository: https://doi.org/10.17605/OSF.IO/XN4G8

Contact: Emma Von Hoene (evonhoen@gmu.edu) and Taylor Anderson (tander6@gmu.edu); Department of Geography and Geoinformation Science, George Mason University

Overview

Public health decision-making—such as identifying health inequalities or targeting interventions—requires population health data at fine geographic scales. However, such ground truth data are often unavailable.

SHAPE (Spatial Health and Population Estimator) introduces an open-source, reproducible, and transparent framework that relies on publicly available data for spatial microsimulation-based synthetic populations and small-area estimation (SAE). Implemented entirely in R, SHAPE uses a hierarchical Iterative Proportional Fitting (IPF) process to integrate individual-level health surveys with spatially aggregated census data. The result is a synthetic population that preserves joint distributions of individual characteristics, enabling spatial aggregation to obtain SAEs. This data allows for fine-scale health analysis and supports downstream applications, such as Agent-Based Modeling, for further exploring disease dynamics or testing policy scenarios.

Methods Summary

SHAPE uses a two-level hierarchical IPF framework:

1. Level 1: Fits BRFSS survey data to ACS demographic marginals to generate synthetic local populations and estimate smoking and obesity prevalence.
2. Level 2: Incorporates Level 1 estimates to model eleven health outcomes, producing fine-scale prevalence estimates.

Health Variables Generated with SHAPE

• Health Behaviors: Smoking, Obesity
• Health Outcomes: Cancer, Asthma, High Blood Pressure, Diabetes, COPD, Arthritis, High Cholesterol, Kidney Disease, Heart Disease, Depression, and Stroke

Repository Structure and Data Organization

This GitHub repository contains materials from two distinct SHAPE studies, organized into separate folders to support transparency, reproducibility, and reuse. A third folder is also included, which holds several R scripts for pre-processing the input data, generating synthetic populations with SHAPE's Level 1 and 2 framework, and post-processing the synthetic populations.

1. SHAPE Evaluation Study

This study systematically evaluated SHAPE-generated SAEs across multiple health behaviors/outcomes, spatial and temporal settings, and spatial scales by comparing them to benchmark datasets (CDC PLACES and BRFSS direct estimates).

• Folder: SHAPE_Evaluation/
• Contents:
  • All input datasets used (BRFSS - Surveys from 2019 and 2021, and ACS - Spatially Aggregated Sociodemographics from 2017-2021 Five Year Estimates)
  • All SHAPE-generated SAE outputs used in the evaluation (Florida - 2019, and New York - 2021- at both county and census tract scales)
  • All evaluation datasets (CDC PLACES - 2019 and 2021 releases at county and census tract levels, and BRFSS direct estimates - county level from New York Department of Health BRFSS 2021, and Florida Health Charts BRFSS 2019)

This folder fully supports the analyses presented in the following SHAPE evaluation manuscript.
Citation: Von Hoene, E., Gupta, A., Kavak, H., Roess, A., & Anderson, T. (2025). Evaluation of A Spatial Microsimulation Framework for Small-Area Estimation of Population Health Outcomes Using the Behavioral Risk Factor Surveillance System. arXiv preprint arXiv:2510.22080.

Link: https://arxiv.org/abs/2510.22080.

2. SHAPE Data Release (2023–2024)

SHAPE was also used to generate national synthetic populations and small-area estimates (SAEs) for recent years, intended as a reusable public data product.

• Folder: SHAPE_Data_Release/
• Years Covered: 2023 and 2024
• Contents:
  • Input data used to generate national synthetic populations (BRFSS - Surveys from 2023 and 2024, and ACS - Spatially Aggregated Sociodemographics from 2019-2023 Five Year Estimates)
  • Output datasets, including:
    • National synthetic populations (individuals aged 18+)*
    • SAEs (county and census tract scale)

These datasets are provided as a standalone data release for downstream research and policy applications.

• Due to large file sizes, both the 2023 and 2024 synthetic population dataset is provided at: https://doi.org/10.17605/OSF.IO/XN4G8.

3. SHAPE R Scripts and Computational Framework

This folder contains all core R scripts used to implement SHAPE, process input data, generate synthetic populations, produce small-area estimates, and perform validation or post-processing analyses. These scripts support both the SHAPE evaluation study and the national SHAPE data releases.

• Folder: R Scripts/

SHAPE development was supported by the National Science Foundation Award No. 2109647 and 2302970.