Researchers in England and Massachusetts have developed an artificial intelligence diagnostic that can predict the likelihood of someone having COVID-19 based on symptoms.

King’s College London, Massachusetts General Hospital and health science company Zoe have developed an AI model that uses data from the COVID Symptom Study app to help predict COVID-19 infection. It compares symptoms and the results of traditional COVID tests to help populations that have limited access to testing.

According to the researchers, more than 3.3 million people around the world have downloaded the app and are reporting daily health status.

The researchers have analyzed data from 2.5 million people in the U.S. and the U.K who have regularly been logging their health in the app. About one-third of the people logged COVID-19 related symptoms. Of the one-third, 18,374 have had a coronavirus test and 7,178 tested positive.

Symptoms associated with COVID-19 were analyzed to see which were most likely to be associated with a positive coronavirus test. The researchers found a wide range of symptoms compared to cold and flu and warned against focusing only on fever and cough. They found that loss of taste and smell were in two-thirds of users who tested positive compared with just over one-fifth who tested negative. The researchers suggest that loss of taste and smell could be a stronger predictor of COVID-19 than fever.

The researchers created a mathematical model that could predict with 80% accuracy whether someone is likely to have COVID-19 based on age, sex and a combination of the symptoms loss of smell or taste, severe or persistent cough, fatigue and skipping meals. They applied the model to the group of over 800,000 app users experiencing symptoms and predicted that just under one-fifth of those who were suck were likely to have COVID-19 at that time.

“Our results suggest that loss of taste or smell is a key early warning sign of COVID-19 infection and should be included in routine screening for the disease. We strongly urge governments and health authorities everywhere to make this information more widely known, and advise anyone experiencing a sudden loss of smell or taste to assume that they are infected and follow local self-isolation guidelines,” King’s College London researcher Tim Spector said in a news release.

Researchers suggest that the AI prediction model with the adoption of the app could help identify people who are likely to be infectious as soon as early symptoms show.