Geo-spatial analysis of racial disparities in 3,952 cardiac arrests: machine learning-augmented multi-site case-control study
Background:
There are no known artificial intelligence (AI)-guided geo-spatial analyses of cardiac arrests.
Methods:
This is a multi-site case-control study of 3,952 cardiac arrest cases from January 1, 2007 to 2016 in the Houston metro area. Geo-spatial analysis was conducted with neural network machine learning-backed multivariable regression of arrest outcomes.
Results:
By zip code there were 16.92 (SD 21.55) arrests on average. Regression analysis showed each additional $10,000 above median household income was associated with a decrease in the total number of cardiac arrests per zip code by 2.86 (95%CI -4.26, -1.46; p<0.001). Zip codes with a median income above $54,600 versus the federal poverty level lowered arrests by 14.62 (p<0.001). For each additional 10 African Americans suffering cardiac arrest in a zip code, the total number of poor neurologic outcomes increased by an average of 8.78 (p<0.001). Geo-spatial maps showed a clock-wise band from north to east to south of more poor neurologic outcomes overlapping with where more African Americans with lower incomes lived (Figure 1).
Conclusions:
This novel geo-spatial AI analysis suggests poorer minorities suffer more adverse cardiac arrest outcomes, potentially providing the opportunity to focus more resources on improving outcomes in less socio-economically advantaged areas.