Precision risk assessment for pediatric hospitalization using address-level data in Cincinnati, Ohio

Area of Science:

• Environmental epidemiology
• Population health science
• Geospatial health analysis

Background:

• Traditional area-level analyses may obscure localized health disparities.
• Integrating socio-environmental data with healthcare data is crucial for understanding pediatric health risks.
• Precision population health requires granular data for targeted interventions.

Purpose of the Study:

• To characterize pediatric hospitalization risk at the residential address level in Cincinnati, Ohio.
• To develop and validate predictive models using socio-environmental and healthcare data.
• To explore the potential of address-level risk prediction for public health.

Main Methods:

• Linked hospitalization data (2016-2022) with parcel-level housing and block-level crime data.
• Utilized U.S. Census and Eviction Lab data, localized to census tracts.
• Employed generalized random forest models to predict hospitalization risk, including a birth-adjusted model.

Main Results:

• Successfully matched 81.5% of hospitalizations to residential addresses.
• Models demonstrated high performance (AUC: 0.98-0.99) in identifying high-risk addresses.
• Key predictors included housing violations, violent crime, and property values.

Conclusions:

• Address-level predictions and multiscale data integration advance precision population health.
• The approach is scalable and privacy-preserving for identifying pediatric health risks.
• Findings support targeted clinical and policy interventions for child health.