Machine Learning Index Identifies Patients at High Risk of COVID-19

Parkland Center for Clinical Innovation (PCCI) and Parkland Health and Hospital System have developed a risk index that generates a COVID-19 risk score for each patient using machine learning and big data.

For more coronavirus updates, visit our resource page, updated twice daily by Xtelligent Healthcare Media.

The Proximity Risk Index leverages geo-mapping and hot-spotting technology to produce a dynamic personal risk score based on a person’s proximity to other confirmed, active COVID-19 cases. To help determine which patients it should reach out to proactively, Parkland created three initiatives using the index to assess individuals’ needs.

“The work PCCI does to connect providers with other community services to improve patients’ social determinants of health is more important now than ever,” said Steve Miff, PhD, PCCI President and CEO. 

The first initiative, called Proximity Index 1.0, was developed and deployed in the spring to identify patients with appointments within the next one to two days who would be coming for an in-person visit to a Parkland facility. The goal was to find patients at risk of having COVID-19 and triage them prior to the appointment to avoid disease spread.

For those suspected to be exposed, Parkland offers them a COVID-19 test and virtual visit. To date, Parkland has screened over 8,000 people using this tool.

The second initiative, Proximity Index 2.0, was implemented the week of September 8, 2020. This tool aimed to address the risk and needs of patients who are not coming into the health system in the next 30 days but have been seen in the past year. The initiative sought to connect directly with more Dallas County residents who are at higher risk of getting COVID-19 and at higher risk of poorer outcomes if they do contract the virus.

These patients would receive a text from Parkland asking questions about their health and activities to help determine potential exposure to, or illness due to, COVID-19. If the patient confirms that they are at risk for COVID-19 exposure, they are provided information about where to get tested and instructions about staying safe.

Additionally, Parkland asks patients with COVID-19-related symptoms if they have medical needs and if so, arrange for a virtual visit for the patient, help obtain medication refills, and meet other healthcare needs.

Parkland also screens patients for food insecurity, partnering with a community-based organization that will deliver food to these potentially ill patients to help them stay well or recover from illness.

The third initiative, Proximity Index 3.0, created geo-mapping of all COVID-19 cases across Dallas County. With the help of PCCI, Parkland identifies trends of where cases are growing, the mobilizes its staff of community health workers along with community leaders to distribute information to those areas about staying safe, monitoring for symptoms and resources for where to get tested and get help.

To date, Parkland has sent text messages to more than 210,000 individuals and made phone calls to over 108,000 people with targeted messaging. Parkland Community Health Workers have had more than 2,800 encounters with at-risk patients, providing social and behavioral support, assistance with medications, COVID-19 education, appointment navigation, and telehealth appointments.

“This project utilizes the key skillsets of Parkland and PCCI — innovation, public service, outreach and community engagement — in service of our patients and the residents of Dallas County. It is greatly extending our reach in combatting COVID-19 effectively, particularly for the most vulnerable in our community,” said Brett Moran, MD, Senior Vice President and Chief Medical Information Officer at Parkland.

PCCI has leveraged data analytics tools to assess patient risk throughout the pandemic. In June, the organization launched a big data dashboard to accurately identify communities at high risk for COVID-19 infection.

“From a community health perspective, these latest PCCI analytics continue to support our ongoing efforts to proactively identify hot-spots, rapidly deploy targeted testing that is accessible locally, inform and educate the community using culturally-sensitive approaches, and align other critical resources to support individuals and their families during these difficult times,” said Philip Huang, MD, Director of Dallas County Health and Human Services.