Information about the novel coronavirus, SARS-CoV-2, and the disease it causes, COVID-19, is accumulating at a stellar rate. However, infections remain difficult to diagnose from symptoms alone, with cases ranging from a mild cough or cold to life-threatening. And with many symptoms similar to cold and flu, it can be different to tell the difference.
Determining which patients do indeed have COVID-19 — and also, which of those patients are likely to develop severe illness — could help doctors treat patients more effectively. Moreover, pinpointing where the next outbreaks are likely to occur will be crucial to halting the pandemic declared earlier this year That’s where artificial intelligence (AI) could step in.
Researchers at King’s College London, Massachusetts General Hospital, and ZOE, a UK-based health science company, have developed a diagnostic tool that can predict whether someone is likely to have COVID-19 based on their symptoms.
The predictive tool uses data from the COVID Symptom Study app, for which there are currently 3.3 billion users reporting their health status on a daily basis, such as whether they feel well or have any new symptoms such as persistent cough, fever, fatigue and loss of taste or smell.
The mathematical model can predict with almost 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, including loss of smell or taste, severe or persistent cough, fatigue and skipping meals, according to a new paper published on 11 May in Nature Medicine (1).
To develop the AI, data were gathered from just under 2.5 million people regularly using the app. Of the one-third who logged symptoms associated with the COVID-19 infection, 18,374 reported having had a test for coronavirus and 7,178 had tested positive. In the entire group of over 800,000 app users experiencing symptoms, the model predicted that just under one-fifth of those who were unwell were likely to have COVID-19.
Combining the AI tool with the widespread adoption of the COVID-19 app could help to identify those who likely to be infectious much sooner, the authors suggest. This could help guide more effective tracking and testing efforts. Two clinical trials are planned in the UK and the US, due to start soon.
Interestingly, loss of taste and smell, called anosmia, was found to be an important indicator of the disease: two-thirds of users testing positive for coronavirus infection reporting this symptom compared with just over a fifth of the participants who tested negative.
“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, who was involved in the study.
“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”.
Other researchers are developing various AI to use in the fight against COVID-19. Two scientists at New York State University are using predictive analytics – a type of AI – to determine when a patient is at risk of developing severe COVID-19 infection.
In another application, a website run by Boston Children’s Hospital called HealthMap is using AI to scan social media, news reports, internet search queries, and other information for signs of disease outbreaks.
While AI might not be able to replace accurate testing, it could help decisionmakers direct testing efforts to suspected hot spots. The question might be whether AI tools can be developed quickly enough to help in the current pandemic.
(1) Menni, C. et al. Real-time tracking of self-reported symptoms to predict potential COVID-19. Nature Medicine (2020). DOI: 10.1038/s41591-020-0916-2