If new AI diagnostic procedures can predict covid-19 infection？

Post date: 2020-05-13

Beijing, May 12 (xinhua) -- scientists from the United Kingdom and the United States have developed an artificial intelligence (AI) diagnostic tool that can predict whether someone is likely to be infected with covid-19 based on symptoms, according to a study published in the journal nature medicine.

The AI model, which USES data from the covid-19 symptoms study app to compare people's symptoms with the results of traditional covid-19 tests to predict whether someone is likely to be infected with covid-19, could help people with limited access to the tests, who are about to start two clinical trials in the UK and us, the researchers said.

More than 3.3 million people worldwide have downloaded the covid-19 symptom study program and use it to report on their health every day.

The researchers analyzed data collected by the app from 2.5 million users in the United States and the United Kingdom who regularly record their health on the app. About a third of those users recorded covid-19-related symptoms. 18,374 reported they had been tested for the coronavirus, and 7,178 of those tested positive.

Using this data, the team looked at which symptoms known to be associated with covid-19 are most likely to be associated with a positive test result.

They found that the symptoms of covid-19 are more varied than those of colds and flu.

They warn that instead of focusing on fever and cough, the loss of taste and smell (anosmia) is of greater concern, as two-thirds of users who test positive for coronavirus infection report symptoms.

This suggests that anosmia is a stronger indicator of covid-19 than fever.

The researchers then created a mathematical model that predicted, with about 80 percent accuracy, whether someone was likely to have covid-19 based on the user's age, gender and four main symptoms -- loss of smell or taste, severe or persistent cough, fatigue and loss of appetite.

When the researchers applied the model to 800,000 symptomatic app users, they predicted that less than one in five (17.42%) of those who were unwell at the time could have been infected with covid-19.

The researchers believe that combining the AI prediction model with the covid-19 symptoms study could identify those at risk of infection when symptoms first appear.

Professor Tim spector, from king's college London, said: "loss of taste or smell is a key warning sign of covid-19 infection and should be included in routine screening for the disease. Anyone who suddenly loses their sense of smell or taste should assume they are infected and comply with local self-quarantine measures."

Source: 《science and technology daily》

The copyright of this article belongs to the original, does not represent the position of this site.

We reprint this article for the purpose of disseminating more information.