A team of developers from Kolozsvár/Cluj-Napoca is working day and night to ease the process of identifying coronavirus infections using machine learning. In order to increase the efficiency of healthcare workers administering the tests, the developers have launched a pilot website designed to pre-screen possible COVID-19 infections in Romania. Users who feel ill can visit the www.covidmonitor.ro website, take the online test, and be notified as to whether they passed the test or need to move forward with real testing.
“Covid Monitor aims to identify patients at high risk from SARS-CoV-2 using artificial intelligence,” project initiator Dan Chiuzbăian, CEO and Founder of QSound SOFT, the company behind the website,
said to Romanian newspaper Adevarul. “There is a two-way data entry: one from patients describing their symptoms, and one from the medical information provided by emergency (triage) centers testing for COVID-19. The algorithm isn’t able to detect the presence of the virus; only an actual coronavirus test (with physical sample) is able to do that,” he added.
What makes this pilot project unique is the way
machine learning is being used to streamline and increase the efficiency of the medical personnel currently under pressure by the rising number of coronavirus infections.
To see whether a patient has been infected with the novel coronavirus or has just caught the flu or maybe pneumonia, emergency centers need to actually test the patient.
Based on the data provided by medical personnel working with the actual tests, the Covid Monitor augments the medical staff’s workflow by eliminating flu cases and low-coronavirus-risk cases, hence reducing transport, time, and energy costs in the healthcare system. As a result, the medical staff can focus on high-risk cases, and everyone gains.
According to Chiuzbăian, there are 200 medical staff registered in the system so far, and they are expecting others to sign up to increase the data points that the algorithm can learn from, which increase the efficiency of the results produced at the end of the test. Registration is currently available only to medical staff.
When the website goes live for regular users – possibly next week – they will complete a test questionnaire and get a result formulated in simple words
such as: “We estimate that there is a 60 percent chance you have a simple cold, a 30 percent chance you have a sore throat, and a 10 percent chance you have some other disease.” Depending on the results, the patient will be instructed to monitor his or her symptoms and retake the test after a while, or, if there are signs that the patient might be infected with COVID-19, he will be instructed to get in touch with a general practitioner.
The project has been supported by the volunteer work of six programmers, physicians, marketing and communication professionals, and legal counsel. The project is backed by the government through the Ministry of Internal Affairs, the Department for Emergency Situations and the Ministry of Health and is currently limited to the website.