Developing a Severity Score to Better Assess COVID-19

John Parkinson

John Parkinson is the senior editor for ContagionLive. Prior to joining MJH Life Sciences in 2020, he has covered a variety of fields and markets including diabetes, oncology, ophthalmology, IT, travel, and local news. You can email him at [email protected]

A diagnostic analyzer was studied looking at 3 biomarkers to predict the level of COVID-19, and could potentially serve as a risk stratification tool.

Assessing the severity of COVID-19 is difficult and in the absence of any diagnostics to gain a better understanding of disease progression makes clinical care challenging.

A new study used an analyzer to look at host proteins to help identify the severity of COVID-19. According to the investigators, it was found to outperform other candidate severity biomarkers including Interleukin 6 (IL-6).

The data was presented at the 23rd Annual Making a Difference in Infectious Disease Meeting 2021 virtual sessions.

The analyzer looked at TNF-related apoptosis-induced ligand (TRAIL), interferon gamma-induced protein-10 (IP-10), and C-reactive protein (CRP). The investigators employed machine learning to integrate TRAIL, IP-10 and CRP levels into a score (0-100). They created 4 To render the model clinically intuitive, 4 score bins were developed.

“In area under the receiver operating characteristic curve (AUC) analysis, the score with AUC 0.86 (95% confidence interval, CI: 0.81-0.91) outperformed other candidate severity biomarkers, including IL-6 (n=139; AUC 0.77 (95%CI: 0.67-0.87); p=0.03),” the investigators reported. “Performance was also assessed by demonstrating a significant trend in likelihood of severe outcome going from low to high score bins (p<0.01). This trend was significant in the sub cohort of severe patients meeting outcome on day of/after blood sampling (n=339; p<0.01).”

The analyzer was developed by Haifa, Israel-based MeMed. Contagion spoke to Tanya Gottlieb, PhD, MBA, vice president, Scientific Affairs, MeMed about the analyzer including how it can help identify what is going on in the immune system and the potential role that machine learning has in clinical care.