AI picks four key indicative symptoms, fever isn’t one...
AI can now tell you whether you have Covid-19 or not, based purely on your symptoms.
Researchers at King’s College London, Massachusetts General Hospital and health science company Zoe developed a diagnostic AI model that analyses data from the Covid-19 Symptom Study app, then compares it to the results of traditional tests for the same patients.
More than 3.3 million people have downloaded the app to report daily on their health status, and track any symptoms that may appear. To build the AI model, the researchers gathered data from 2.5 million users in the UK and US who have been keeping their symptoms up to date in the app. Around a third of these people had logged symptoms associated with Covid-19, and of these, 8,374 reported having had a test for coronavirus, with 7,178 people testing positive.
The resulting model can predict whether a person has Covid-19 with almost 80% accuracy. It uses several different symptoms as well as the person’s age and sex. Loss of taste and smell was found to be a particularly good indicator of Covid-19, but the model also needs data about whether the person has a severe or persistent cough, whether they are experiencing fatigue, and whether they are skipping meals. It is interesting to note that fever, commonly thought to be an indicator of Covid-19 infection, is not one of the symptoms considered indicative by the model.
“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,” said Professor Tim Spector from King’s College London. “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.”
The researchers behind this study suggest that widespread adoption of the Symptom Study app combined with the AI model could be used to help identify patients who are likely to have contracted Covid-19 without the need for a clinical test. This could help patients decide when to self-isolate, and augment tracking and testing efforts. The AI-powered diagnostic may also be able to help populations who do not have access to Covid-19 testing.
The results of this study are published in Nature Medicine. Clinical trials in the UK and US will begin shortly.