RANDOM FOREST AND SHAP ANALYSIS ON SOCIAL VULNERABILITY AND EDENTULISM AMONG US ADULTS 65 AND OLDER

Sep 29, 2026

Upload by pop out

Approximately 17.3% of U.S. adults aged 65 and older experience complete edentulism. The study integrates CDC Places (2022) and Social Vulnerability Index (SVI) (2022) data across more than 3,000 counties, emphasizing interactions among social factors. Random Forest and SHAP are used to identify key drivers and quantify interaction effects. Poverty rate is the strongest predictor, with education and internet inaccessibility following, and poverty–no-high-school-diploma interaction showing the greatest impact.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages2

Reading Time5 min

AccessFree to download

Abstract

Data sources and study scope

Modeling approach (Random Forest and SHAP)

Key predictors and interaction effects

Thresholds and intervention implications

Innovation in Aging, 2025, Vol. 9, No. S2 1251 Abstract citation ID: igaf122.4181 RANDOM FOREST AND SHAP ANALYSIS ON SOCIAL VULNERABILITY AND EDENTULISM AMONG US ADULTS 65 AND OLDER Jinbo Niu1, Joonhyeog Park2, and Tamara Cadet2, 1. University of Southern California, Los Angeles, California, United States, 2. University of Pennsylvania, Philadelphia, Pennsylvania, United States Approximately 17.3% of adults aged 65 and older in the U.S. face complete edentulism. Identifying the social factors that contribute to edentulism and their interactions is crucial for creating predictive models and effective interventions. However, most studies ignored the interactions among social factors on the edentulism prevalence. Using the concept of social vulnerability, we integrated the latest CDC Places data (2022) and Social Vulnerability Index (SVI) data (2022) to analyze the prevalence of edentulism among adults aged 65 and older across over 3,000 U.S. counties. Random Forest (RF) and SHapley Additive exPlanations (SHAP) were applied to identify the main driving 1252 Innovation in Aging, 2025, Vol. 9, No. S2 factors of edentulism and characterize their interaction effects. Among the factors considered, the poverty rate emerged as the most influential in predicting edentulism prevalence, followed by the no-high-school-diploma rate and the internet inaccessibility rate. The interaction between poverty rate and no-high-school-diploma rate had the strongest effect on edentulism prevalence. On average, the interactions between social vulnerability factors accounted for 123.08% of the direct influence on edentulism prevalence. The analysis also revealed critical thresholds where the combined impact of these factors shifts to either exacerbate or moderate edentulism prevalence. Social vulnerabilities do not operate in isolation, and their interactions can exacerbate or moderate edentulism among older adults. Accounting for interaction effects and thresholds can refine prediction modeling and guide integrated interventions addressing poverty, education, and internet access to better target high-risk communities.

FAQs

What was the study goal regarding edentulism and social vulnerability?

To identify social factors contributing to edentulism among U.S. adults aged 65 and older and to analyze how interactions among these factors affect prevalence.

Which data sources and geographic coverage were used?

The analysis combined CDC Places (2022) and SVI (2022) data across over 3,000 U.S. counties.

How were Random Forest and SHAP applied in the study?

Random Forest and SHAP were used to determine the main driving factors of edentulism and to characterize their interaction effects, including the strength of combined influences.

Sep 29, 2026

Upload by pop out