Since the earliest days of the COVID-19 pandemic, one of many largest challenges for well being programs has been to acquire an understanding of the neighborhood unfold of this virus and to decide how seemingly is it that an individual strolling by way of the doorways of a facility is at the next risk of being COVID-19 optimistic.
Without sufficient entry to testing knowledge, well being programs early-on have been typically pressured to depend on people to reply questions similar to whether or not they had traveled to sure high-risk areas. Even that unreliable technique of assessing risk began changing into meaningless as local people unfold took maintain.
Parkland Health & Hospital System, the security web well being system for Dallas County, Texas, and PCCI, a Dallas-based non-profit with experience within the sensible functions of superior knowledge science and social determinants of well being, had a greater thought.
Community unfold of an infectious illness is made attainable by way of bodily proximity and density of energetic carriers and non-infected people. Thus, to perceive the risk of a person contracting the illness (exposure risk), it was crucial to assess their proximity to confirmed COVID-19 instances elegant on their deal with and inhabitants density of these places.
If an “exposure risk” index may very well be created, then Parkland may use it to decrease exposure for his or her sufferers and well being employees and present focused academic outreach in extremely susceptible zip codes.
PCCI’s knowledge science and medical crew labored diligently in collaboration with the Parkland Informatics crew to develop an modern machine learning pushed predictive mannequin referred to as Proximity Index. Proximity Index predicts for a person’s COVID-19 exposure risk, elegant on their proximity to check optimistic instances and the inhabitants density.
This mannequin was put into motion at Parkland by way of PCCI’s cloud-based superior analytics and machine learning platform referred to as Isthmus. PCCI’s machine learning engineering crew generated geospatial evaluation for the mannequin and, with assist from the Parkland IT crew, built-in it with their digital well being file system.
Since April 22, Parkland’s inhabitants well being crew has utilized the Proximity Index for 4 key system-wide initiatives to triage greater than 100,000 affected person encounters and to assess wants, proactively:
- Patients most at risk, with appointments in 1-2 days, have been screened forward of their go to to stop unfold inside hospital
- Patients recognized as susceptible, have been provided further medical (i.e. digital go to, medicine refill help) and social assist
- Communities, by zip-code, most at-risk have been despatched focused messaging and targeted outreach on COVID-19 prevention, staying secure, monitoring for signs, and sources for the place to get examined and medical assist.
- High exposure risk sufferers who had an appointment at considered one of Parkland’s neighborhood clinics within the subsequent couple of days have been provided a telehealth appointment as an alternative of a bodily appointment if that was applicable elegant on the kind of appointment.
In the long run, PCCI is planning on providing Proximity Index to different organizations in the neighborhood – colleges, employers, and so on., in addition to to people to present them with an information pushed device to assist in resolution making round reopening the economic system and society in a secure, considerate method.
Many groups throughout the Parkland household collaborated on this undertaking, together with the IT crew led by Brett Moran, MD, Senior Vice President, Associate Chief Medical Officer and Chief Medical Information Officer at Parkland Health and Hospital System.