AI Powers a Potential App for At-Home Coronavirus Risk Assessment
As reports about COVID-19 continue to dominate the news cycle, researchers and healthcare professionals are working around the clock to find ways to contain and mitigate the global impact of the novel coronavirus. Scientists have already started trials of a recently developed coronavirus vaccine on humans, while other researchers work to create diagnostic tools that will help accelerate the rate of testing in the United States. With the varied patchwork of responses currently being implemented on both the federal and state level, it seems that a tech-oriented approach may help immensely to fill in the gaps, especially when it comes to the inadequate levels of coronavirus testing within the US.
One such group from Augusta University’s Medical College of Georgia is now working on an AI-powered app that allows individuals to assess their risk of contracting the virus, from the comfort of their home. Besides offering users a quick method of evaluation and guiding those who are at greatest risk to the nearest testing facility, the app would also offer local health authorities information that’s being gathered in real-time, so that they can develop more targeted treatment and prevention measures.
According to the team, whose preliminary paper was recently published in the journal of Infection Control & Hospital Epidemiology, the idea was to help identify people who are at highest risk of contracting the virus, and to accelerate the process of screening and identifying those who are actually at risk, so that only those who actually need it can access limited medical resources, and therefore to help reduce the severity of the impact of the coronavirus overall.
“This is an AI model-based framework for identifying people who are at risk of COVID-19 and people who are not at risk,” said Dr. Arni S.R. Srinivasa Rao, a health sciences professor and director of the Laboratory for Theory and Mathematical Modeling at Medical College of Georgia’s Division of Infectious Diseases. “The aim is to make it home-based assessment, so that only those who are in real need can use the hospital facilities. We saw in several countries with high COVID-19 numbers that the hospitals do not have enough manpower and beds to take care of the patients. So the health-related workforce and facilities need to be spared for those who are in real need.”
According to Rao, the app will use a “reactive” type of machine intelligence, which will help experts collect and process large amounts of data, in order to assist them in making decisions based on the status of COVID-19 risk in any given region.
Demographic information such as age, gender, race will be inputted by the user, in addition to important details such as whether the individual has traveled to outbreak “hotspots” in countries such as China, Iran or Italy within the last 14 days. The app will also gather information on possible symptoms — like fever, cough, shortness of breath, fatigue, phlegm production, headache, diarrhea and pneumonia — for the individual user as well as for those that live in close contact, but who might not be able to complete their own medical questionnaire.
Using this data, the app’s algorithm can then determine the user’s level of risk — none, low, moderate or high risk — and then automatically notify the nearest testing center to either schedule a health check either on its premises or remotely, or to send a mobile testing team to perform one.
Currently, Rao and paper co-author Dr. Jose Vazquez, a professor of medicine and chief of the MCG Division of Infectious Diseases, are now collaborating with developers to bring a finished version of the app online within the next couple of weeks. It will be offered as a free service on Augusta University’s website, and users can download it both as an iOS or Android app. The process is going forward altogether very quickly, acknowledges Rao, and if needed, it still might take some extra time to obtain approval from the FDA for such a mobile medical application.
“As a mathematical modeler, I would have also spent time in modeling the spread, but instead, I chose to build the AI model quickly and developed a theory that can identify possible cases of COVID-19, based on guidelines set by CDC Atlanta which were available publicly,” explained Rao. “I thought the key challenge is to identify individuals who are at risk because the virus is novel.”
The team’s ultimate aim is to increase accessibility, accelerate identification of at-risk populations, reduce costs and relieve some of the time pressures and shortages that are currently swamping medical professionals as the crisis grows. In the future, the pair envisions that the app could be adapted for other pandemics so that experts can better pinpoint regions or demographics that are under threat, allowing them to then distribute healthcare resources accordingly, or to quickly pre-screen people prior to large gatherings.
“We are trying to decrease the exposure of people who are sick to people who are not sick,” said Vazquez. “We also want to ensure that people who are infected get a definitive diagnosis and get the supportive care they may need.”
Read the team’s paper here.
Images: Augusta University