07 Harnessing Machine Learning to Measure Access to Health Care

Objectives

Are you aware of how your community’s health requirements are being addressed? Ensuring equitable health care accessibility endows all individuals with equal opportunities for care, regardless of their socioeconomic variances. Prior studies have shown that economically disadvantaged communities frequently experience elevated breast cancer mortality rates within the US.

Materials and Methods

We modeled breast cancer mortality rates using a Gaussian linear regression model. The predictors variables were mammogram screening, smoking, and diabetes rates as well as Area Deprivation Index (ADI), rural/urban continuum codes, and racial population data.

Results

Our findings indicated that breast cancer mortality risk is the highest among African American individuals. Women in more rural and higher ADI counties also had much higher mortality rates than women in more urban and lower ADI counties, respectively. Counties with greater rates of smoking and diabetes also were associated with higher breast mortality rates. Higher mammogram rates were associated with lower mortality rates.

Conclusions

By employing sophisticated machine learning techniques, we predicted breast cancer mortality rate across the United States. With these results coupled with other derived metrics, we identified key patterns and drivers of inequity in breast cancer outcomes. We aim to provide more effective support to underserved cancer-vulnerable communities through diverse outreach initiatives.