Machine Learning-Based Rapid Diagnostic Reads Immune System to Predict COVID-19 Severity Risk

A rapid diagnostic that reads the immune system to predict severe respiratory failure risk in COVID-19 patients is being developed to help physicians make better hospital admission and resourcing decisions for COVID-19 patients at hospital presentation.

Inflammatix (Burlingame, CA, USA) has been awarded USD 1.1 million by the Defense Advanced Research Projects Agency (DARPA) for further development of the rapid diagnostic named CoVerity COVID-19 Severity Test. The Inflammatix approach – known as host-response diagnostics – rapidly reads the immune system using multiple mRNA biomarkers and a machine learning algorithm. The company is developing other host-response diagnostic tests that identify the presence and type of infection (viral or bacterial), in addition to predicting the risk of severe disease, to enable physicians to make more informed decisions for patients with acute infection and sepsis.


The company’s host-response diagnostic approach for predicting COVID-19 severity risk was shown to be superior to clinical biomarkers, including IL-6, in a new study presented recently at the 2020 European Society of Clinical Microbiology and Infection Diseases (ESCMID) Conference on Coronavirus Disease (ECCVID). In a prospective study of 97 patients with PCR-confirmed SARS-CoV-2 pneumonia and blood drawn on the day of hospital admission, CoVerity demonstrated an area under the receiver operating characteristic curve (AUROC) of 0.88 (95% CI 0.81-0.95) for identifying patients who developed respiratory failure or died, independent of age, while IL-6 had an AUROC of 0.73 (95% CI 0.62 - 0.85). The new classifier had the highest accuracy among all single biomarkers tested, including IL-6, procalcitonin, C-reactive protein, lactate, and SuPAR.

“While major progress has been made in developing rapid platforms to diagnose SARS-CoV-2 infection, predicting severity in COVID-19 patients remains an unmet medical need,” said Evangelos J. Giamarellos-Bourboulis, MD, Professor of Internal Medicine and Infectious Diseases at ATTIKON University General Hospital in Athens, Greece, Chairman of the European Sepsis Alliance, President of the European Shock Society, and lead investigator for the study. “In this study, the host-response approach demonstrated very high accuracy for identifying severe disease in COVID-19 patients and outperformed clinical markers for risk stratification. Existing tools have shown limited accuracy in enabling us to confidently identify high-risk patients early who need close monitoring or discharge non-severe patients to recover at home.”

“We are grateful that DARPA has recognized the promise of our host-response approach to benefit COVID-19 patients and caregivers, and we look forward to accelerating development and availability of our CoVerity COVID-19 Severity Test as a result of this agreement,” said Inflammatix CEO and Cofounder Tim Sweeney, MD, PhD. “The 5-mRNA classifier for CoVerity was developed on a training set of more than 20 clinical studies and we intend to translate it into a rapid assay that can be used as a clinical tool to help triage patients after diagnosis with COVID-19. Improved triage has the potential to reduce morbidity and mortality while enabling hospitals to allocate resources more effectively.”

Related Links:
Inflammatix