Knowing how saturated the triaging sector is, we were hesitant to pursue the idea. However, after speaking to experts in public health and front-line workers, we heard that these triaging platforms were not inclusive towards dense, low-income population centers. The primary reason being that many of the people in these population centers do not have the proper insurance or Medicaid to get tested. So, even if they use a triaging system to get diagnosed, there was no way they could actually get tested. Our team is based in St. Louis, a city that is already struggling to provide healthcare towards low-income populates. We, therefore, decided to not only construct a technical solution but also revise the operational workflow of triaging to make it more inclusive. The primary way we did this was partnering with local testing centers and seeing whether we could create a risk assessment test that would allow anyone that uses our platform and receives a high-risk diagnosis to get tested at our partnering testing center. This way, Compass, our triaging system, not only addresses whether people should get tested but also, how they can get tested.

First, the patient uses our SMS system and answers specific questions that can assess the risk of having COVID-19. Once the risk assessment is complete, they receive their risk assessment and whether they should visit the clinic, the locations/phone numbers of our partner testing centers, and a four-digit authentification code that can be used by the testing center to validate that the patient used our system. The patient can then visit the clinic and give them a call when they are outside. The tester will validate the profile over the phone and then give the patient-specific PPE to prevent the spread of the virus in the clinic. The patient can then either get tested in the car or at the testing center. We are specifically solving the lack of triage testing due to a lack of insurance and Medicaid by partnering and working with local testing centers to provide a set of guidelines that indisputably allows a patient to get testing. In this sense, this is not just a technical solution, but rather an operational solution that addresses a systemic public health concern.

We started by working with testing centers and physicians to construct a set of questions that can accurately assess risk specifically in low-income populations. We used these questions to construct an SMS flow using Twilio. We parse the patient's responses using Twilio's integrated NLP - we use these responses and a risk algorithm to assess whether a patient is at risk of having COVID-19. We are using Microsoft's Azure SQL Database to securely store the patient responses - this data can be accessed by a portal on the side of the testing center when a patient comes in to get tested. Our SMS platform also provides a random four-digit code to identify the patient and the profile that is stored in our database. This prevents any personal health information violations because a randomized four-digit code cannot be traced back to a person. Now, a patient has been given a unique identifier that our testing center also has, and they have also been given the locations of our testing centers they should visit. Finally, we have an input stream that allows either the patient or the test provider to enter the final result of the patient's COVID-19 test - this response is used to assess the accuracy of our system.

One of the biggest challenges we ran into was how to assess the risk someone is at for having COVID. The symptoms for COVID are rather sporadic and there is no one clear set of symptoms that automatically qualifies as a positive test. We addressed the problem by working with physicians and testing centers to create the most comprehensive questionnaire set. We then implemented these questions through our SMS platform.