COVID-19 Resource Center

COVID-19 Resource Center

Disclosures: The authors report no relevant financial disclosures.
November 10, 2020
1 min read

AI-generated score could predict prognosis of outpatients with COVID-19

Disclosures: The authors report no relevant financial disclosures.
You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact

A novel artificial intelligence-generated score can predict the prognosis of patients who present with COVID-19 in the outpatient setting, according to results from a single-center study published in The Journal of Infectious Diseases.

The prediction model, which can be implemented in electronic health records, was created by investigators at Massachusetts General Hospital and validated prospectively. The investigators named it the COVID-19 acuity (CoVA) score.

“Testing the model prospectively helped us to verify that the CoVA score actually works when it sees 'new' patients, in the real world," Haoqi Sun, PhD, an investigator in the department of neurology and a research faculty member in the Massachusetts General Hospital Clinical Data Animation Center, said in a press release.

Sun and colleagues created the CoVA score based on data from 9,381 adult outpatients treated in the hospital’s EDs or respiratory illness clinics (development group), and then prospectively tested it on 2,205 patients treated in the same clinics (prospective group). The primary outcome was the occurrence of an adverse event within 7 days following an outpatient medical encounter, including critical illness (ICU visit or ventilation), hospitalization or death.

The CoVA score had “excellent” discrimination and calibration in the development group in regard to hospitalization (area under the curve = 0.8; 95% CI, 0.79-0.81), critical illness (AUC = 0.82; 95% CI, 0.8-0.83) and death (AUC = 0.87; 95% CI, 0.83-0.91). The CoVA score’s performance was similar in the prospective group for hospitalization (AUC = 0.76; 95% CI, 0.73-0.78), critical illness (AUC = 0.79; 95% CI, 0.75-0.82) and death (AUC = 0.93; 95% CI, 0.86-0.98).

"The large sample size helped ensure that the machine learning model was able to learn which of the many different pieces of data available allow reliable predictions about the course of COVID-19 infection," M. Brandon Westover, MD, PhD, investigator in the department of neurology and director of data science at the Massachusetts General Hospital McCance Center for Brain Health, said in the release.

“Further research is needed to see if these results hold more broadly and as the epidemic matures,” the authors wrote. “In conclusion, CoVA is a well-calibrated, discriminative, prospectively validated and interpretable score that estimates the risk for adverse events among outpatients presenting with possible COVID-19 infection.”


Press Release

Sun H, et al. J Infect Dis. 2020;doi:10.1093/infdis/jiaa663.