Abstract C008: Exploring socioeconomic and racial influences on breast cancer comorbidity in the Memphis metropolitan area: A geospatial and machine learning analysis

Abstract Introduction Memphis, Tennessee, ranks among the top U.S. cities for breast cancer mortality, especially among African American women. Breast cancer presents a significant public health challenge, exacerbated by various comorbid conditions that complicate patient outcomes. The Charlson Comorbidity Index (CCI) predicts ten-year mortality risk based on comorbid conditions, with higher scores indicating greater risk. Despite its importance, localized data on factors influencing CCI scores among breast cancer patients is limited. This study aims to identify spatial hotspots of breast cancer incidence and comorbidity index in Shelby County census tracts and use machine learning to determine individual and neighborhood-level factors contributing to higher comorbidity among breast cancer patients. Methods Clinical data on 12,409 breast cancer patients from 2014 to 2021 were obtained from the Research Enterprise Data Warehouse (rEDW). This dataset included anonymized information on age, census tract geoid, race, Charlson Comorbidity Index (CCI), mammography screening, smoking, alcohol, and substance use. Spatial clusters (hotspots) of annual breast cancer incidence and average CCI in each tract were identified using Getis-Ord Gi* statistics in ArcGIS Pro. These clusters were then overlaid with the distribution of racial segregation and SVI for comparative analysis. The data were divided into training (75%) and testing (25%) sets and stratified by age groups for analysis. A Random Forest (RF) model was implemented using the “randomForest” package in R, with hyperparameters optimized through grid search to minimize the root mean squared error (RMSE). The RF algorithm ranked variable importance by summing decreases in impurity, effectively identifying key predictors. Results Among the 12,409 breast cancer cases recorded during the study period, 7,782 were from Shelby County, with the remainder from neighboring counties. Hotspots of annual breast cancer incidence rates were observed in the western and southern parts of the county, while CCI clusters were primarily concentrated in the western areas, including downtown Memphis. When compared with demographic maps, the hotspots predominantly overlapped with census tracts that have a higher proportion of Black populations. The RF model identified the social vulnerability index (SVI), racial segregation, and race as the most important factors across all age groups (0-40, 40-65, and 65+). Conclusion Our study underscores the critical role of socioeconomic factors and racial segregation in shaping the comorbidity index of breast cancer patients, as evidenced by both area-level and individual-level analyses. The spatial disparities in the comorbidity index, particularly pronounced in areas with racially segregated Black populations and high SVI scores, highlight the need for targeted interventions. By focusing resources on these areas, we can potentially improve patient survival rates and enhance the effectiveness of breast cancer treatment and health and wellness management strategies. Citation Format: Soheil Hashtarkhani, Shelly White-Means, Sam Li, Lokesh Chinthala, Fekede Kumsa, Cindy Kaye Lemon, Lluvia Chipman, Jill Dapremont, Tronlyn Thompson, Arash Shaban-Nejad. Exploring socioeconomic and racial influences on breast cancer comorbidity in the Memphis metropolitan area: A geospatial and machine learning analysis abstract. In: Proceedings of the 17th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2024 Sep 21-24; Los Angeles, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2024;33(9 Suppl):Abstract nr C008.