Researchers at King’s College London, Massachusetts General Hospital and health science company ZOE have developed an artificial intelligence diagnostic that can predict whether someone is likely to have COVID-19 based on their symptoms.
Their findings have been published today in Nature Medicine.
The AI model uses data from the COVID Symptom Study app to predict COVID-19 infection, by comparing people’s symptoms and the results of traditional COVID tests.
Researchers say this may provide help for populations where access to testing is limited.
Two clinical trials in the UK and the US are due to start shortly.
The study noted that more than 3.3 million people globally have downloaded the COVID Symptom Study app, and are using it to report daily on their health status, whether they feel well, or have any new symptoms such as persistent cough, fever, fatigue, and loss of taste or smell.
According to the scientists, the model may provide help for populations where access to testing is limited.
In the current study, the researchers analysed data gathered from just under 2.5 million people in the UK and US who had been regularly logging their health status in the app. The scientists said around a third of the users had logged symptoms associated with COVID-19.
Of these, more than 18,000 reported having had a test for coronavirus, with 7,178 people testing positive, the study reported.
The researchers attempted to understand which symptoms linked to COVID-19 were most likely to be associated with a positive test.
They found a wide range of symptoms compared to cold and flu, and warned against focusing only on fever and cough.
Loss of taste and smell was particularly striking, the scientists said, with two thirds of users testing positive for coronavirus infection reporting these symptoms, compared with just over a fifth of the participants who tested negative.
Based on the findings, the researchers suggested that the loss of sense of smell (anosmia) is a stronger predictor of COVID-19 than fever. Using a new mathematical model which they created, the scientists then predicted with nearly 80 per cent accuracy whether an individual is likely to have COVID-19 based on their age, sex, and a combination of four key symptoms.
These symptoms were disruption of sense of smell or taste, severe or persistent cough, fatigue, and the loss of appetite, the study noted.
By applying this model to the entire group of over 8,00,000 app users experiencing symptoms, the scientists predicted that just under a fifth of those who were unwell were likely to have COVID-19 at that time.
Combining this AI prediction with widespread adoption of the app can help in identifying those who are likely to be infectious as soon as the earliest symptoms start to appear, the scientists said.
They added that this would help in focussing tracking and testing efforts where they are most needed. Citing the limitations of the study, the researchers said the prediction is based on self-reported nature of the included data, which they said cannot replace physiological assessments of smell and taste function, or testing people’s samples for SARS-CoV-2 genetic material.
Another drawback in the study, cited by the researchers, was that the volunteers using the app are a self-selected group who might not be fully representative of the general population.
“We strongly urge governments and health authorities everywhere to make this information more widely known, and advise anyone experiencing sudden loss of smell or taste to assume that they are infected and follow local self-isolation guidelines,” said study co-author Tim Spector from King’s College London.