Knowing whether a person has a fever or not is a good measure to identify possible patients infected with the new coronavirus (SARS-CoV-2). This is because, in addition to coughing and shortness of breath, the change in temperature (above 37.8 °) is one of the main symptoms of COVID-19. With that in mind, two Brazilian startups developed Fevver, an Artificial Intelligence (AI) device that identifies patients with fever.
Today, individuals who arrive at Hospital Israelita Albert Einstein, in São Paulo, already have their temperatures measured automatically, at a distance, by this computer vision system installed on a totem at the reception of the place.
Named after the word fever, in English, the equipment has a thermographic camera and facial recognition algorithms, which scan people’s faces and measure their respective temperatures in an automated way.
When AI technology detects that a patient has a fever, it sends an alert to the hospital’s nursing staff, which initiates a quick screening and isolation protocol to prevent that person from transmitting the virus to other patients in the hospital environment.
How does Fevver work?
This system was developed by two Brazilian startups, Hoobox and Radsquare, incubated by the Albert Einstein Hospital itself. Hoobox is also supported by the Fapesp Innovative Research in Small Business Program (Pipe), of the São Paulo Research Foundation.
The developed tool mixes different technologies, such as face detection, developed by Hoobox, in its first stage. Then, to detect the rise in body temperature, the technology that has already identified the faces of possible patients, extracts points from the corners around the eyes with high precision and discards physiological noises, such as sweat.
Through a technology for thermal detection of infrared energy radiation (thermography), developed by Radsquare, the temperature of the corners of the eyes, where the tear ducts are located, is measured.
“How are structures without epidermal coverage [de pele], have relatively stable humidity and are vascularly very close to the brain, where body thermal control is performed, the tear ducts are the ideal places to assess body temperature by thermography ”, explains Felipe Brunetto Tancredi, CSO of Radsquare to Fapesp.
In case the technology detects that a patient has a fever, the Fevver system takes a photo and generates a notification to the nursing staff or the hospital reception, which can easily identify the patient.
Extended use of Fevver
“This technology is extremely useful for screening very quickly and directing people who have a fever and eventually with COVID-19 to a suitable location. This increases safety not only for patients, but for hospital staff, ”says José Cláudio Cyrineu Terra, director of innovation at Hospital Albert Einstein.
In addition to the hospital and other sectors of the private network, the researchers’ idea is that this system can be used in field hospitals, which have started operating in the last few days in different regions of the country, in addition to the Unified Health System (SUS) .