How science and data can help Texas vaccinate those most at risk

Nearly two months into the COVID-19 vaccine distribution process, Black and Hispanic Texans have received just 8 and 15 percent of available doses, respectively, despite accounting for 19 and 44 percent of cases statewide. This discrepancy shows that decisions about who should get the vaccine first have not been based on actual risk. To correct course, state public health officials need a scientific and data-based system for identifying individuals whose medical history or environment make them more susceptible to the worst outcomes when exposed to the virus – and these people should be first in line for shots.     

As the distribution effort expands, it must address the glaring disparities that the pandemic has exposed, not make them worse. Communities of color in Texas have been devastated by the pandemic. Specifically, death rates among Hispanics and African Americans have been disproportionately high. Hispanics ages 25 to 64, for example, are dying of COVID-19 at a rate more than four times as high as that of non-Hispanic whites. For African Americans in that age group, the COVID-19 death rate is more than twice the rate of whites.  

With more than 8 million Texans eligible for vaccination at a time when COVID-19 cases and hospitalizations remain at an all-time high, top state health officials, who rebuked a strategy to prioritize at- risk minorities in Dallas just last week, must urgently improve the existing distribution plan.  

To do so effectively and equitably, officials should hone the power of science and data to precisely identify the most vulnerable individuals. This means taking into account both the clinical and environmental factors — everything from air quality to housing density and access to fresh food — impacting an individual’s risk of ending up requiring hospitalization or intensive care after contracting the virus. For reaching communities of color in particular, considering clinical risk alongside social determinants of health is critical for determining who should be at the front of the vaccine queue, as underlying health conditions such as obesity, diabetes, or hypertension –– all of which are prevalent among minority communities, but can often be overlooked ––  can uniquely put them at higher risk of death if infected by the virus. 

Machine learning tools can allow officials to quickly and accurately make sense of all of this data. We’ve begun to see these tools in action during a recent pilot project between my data-science company Cogitativo and the U.S. Department of Health of Human Services in December. Through this innovative private-public partnership, I saw firsthand the importance of leveraging data science expertise to build machine learning models that can help government and industry partners lead better outcomes for patients. 

As the vaccines were going through the approval process, Cogitativo analyzed troves of demographic, clinical, and social determinants of health data — in addition to peer-reviewed COVID-19 medical literature — for more than 20 million Americans to identify those populations at greatest risk of illness if exposed to COVID-19. The results helped HHS identify thousands of patients who would have otherwise faced barriers to receiving the vaccine without targeted intervention. 

When it comes to developing a successful vaccination queue, Texas has an opportunity to take a smarter approach and turn to systems rooted in science and data to identify and protect the states most at-risk minority populations. While the vaccine offers hope that we are on the road to recovery, we need to make sure we are able to solve the urgent questions around getting it to those who need it most, and machine learning is poised to help give us those answers throughout Texas. 

Velasquez is the co-founder and CEO of Cogitativo, a data science company based in Berkeley, Calif.

This article originally appeared on Austin American-Statesman: How science and data can help Texas vaccinate those most at risk

Continue Reading