AI diagnostic predicts COVID-19 without testing

Editor's Note

In this study, researchers at the Massachusetts General Hospital, Boston; King’s College, London; and health science company Zoe Global, Accra, Ghana; developed an artificial intelligence (AI) diagnostic that can predict whether someone is likely to have COVID-19 based on their symptoms.

The researchers analyzed data from 2,618,862 people (2,450,569 UK, 168,293 US) who had regularly logged their health status into a smartphone-based app, around a third of whom had logged symptoms linked to COVID-19. An AI mathematical model was applied to the data, and it predicted that 140,312 (17.42%) were likely to have COVID-19.

The AI model predicted with nearly 80% accuracy whether an individual was likely to have COVID-19 based on their age, gender, and a combination of four key symptoms:

• loss of smell or taste
• severe or persistent cough
• fatigue
• skipping meals.

The researchers suggest that combining the AI prediction model with widespread adoption of the app could help identify those who are likely to be infectious as soon as symptoms start to appear.