Now Reading
How We Can Use Technology to Help Identify Higher Risk COVID-19 Patients Early

How We Can Use Technology to Help Identify Higher Risk COVID-19 Patients Early

COVID-19 continues to be a global challenge, with over 7.3 million cases and 416,000 deaths (at the time of writing). Some 120,000 new cases are being detected every day, stretching medical facilities in even the most developed nations.

As India emerged from a 75-day nationwide lockdown on June 1, the stark reality of a surging outbreak stares at us. During the lockdown, cumulative cases and deaths in India rose from 657 to 257,486 (392-times) and from 12 to 7,207 (600-times), respectively. On average, India is now adding about 10,000 cases and 300 deaths daily.

Models have predicted that by the end of June 2020, India would need about 57,000 ICU beds (from 38,000 to 76,000). By the end of July, this number is expected to reach about 141,000 ICU beds (94,000-188,000). There are an estimated 95,000 ICU beds in India, but these are not distributed evenly across the country. Physicians are already facing the onerous task of deciding who needs to be in a hospital and who can recover at home. For those in the hospital, the demand for ICU beds may soon outstrip their availability.

Can science help here? That is, can there be a simple blood test to separate infected people into those likely to develop mild disease and who can recover at home, and those likely to develop moderate to severe COVID-19 and who are likely to require hospital or ICU care?

In a paper scheduled to appear in the July 7 issue of the journal Cell, researchers from Westlake University in China show that this may be possible.

They asked what distinguishes about 80% of the people who have tested positive for the novel coronavirus who recover with little or no medical intervention from the remaining 20%, who develop serious illness. They analysed blood samples from 53 healthy people and 46 COVID-19 patients, of which 21 had severe disease with reduced oxygen levels and respiratory distress, for differences in their protein and metabolite signatures by mass spectrometry.

Of the 894 proteins and 941 metabolites identified from the samples, 93 proteins and 204 metabolites correlated with disease severity. When the researchers analysed them further, these mapped to three main biological processes. Two were related to the immune system, including early immune responses and the scavenging function of macrophages (a type of blood cell). The third related to the function of platelets, a blood cell required for clotting. Some studies have already implicated pulmonary thrombosis, or progressive clotting of blood in the lungs, in a large fraction of COVID-19 deaths.

The researchers then used machine-learning approaches to determine whether a set of these proteins and metabolites could be used to predict disease severity. Such patient stratification was possible with a set of 29 blood factors (22 proteins and seven metabolites) with 94% accuracy in the training set. In a validation set of 10 independent patients, the system picked seven accurately.

“It is a good but preliminary study showing the potential of profiling proteins and metabolites in COVID-19 patients,” said Akhilesh Pandey, a clinical researcher and an expert in mass spectrometry. He is a professor of laboratory medicine and pathology at the Mayo Clinic, US.

In another report published on May 14 in Nature Machine Intelligence, scientists in Wuhan, China, used a database of blood tests and clinical outcomes of 485 COVID-19 patients to identify predictive markers of disease mortality. The machine learning tools selected three biomarkers: lactic dehydrogenase (LDH), lymphocytes1 and high-sensitivity C-reactive proteins (hs-CRP) that could predict the mortality of individual patients more than 10 days in advance, with 90% accuracy. This allows patient-prioritisation in a clinical setting, which would save lives at a time when hospitals are over-burdened.

With a grant from the DBT/Wellcome Trust India Alliance, Pandey established the Centre for Molecular Medicine at the National Institute for Mental Health and Neurological Sciences, Bengaluru, which is also a designated COVID-19 testing centre. (Editor’s note: The author is the CEO of the DBT/Wellcome Trust India Alliance.) “Some people from Bengaluru are training here with me at the Mayo Clinic in ongoing research on COVID-19,” Pandey said. He hopes to replicate these studies on Indian patients as well. “COVID-19 is here to stay and good studies are needed from all over the world.”

Several studies have reported clinical parameters related to COVID-19 outcomes. A meta-analysis of four studies showed that increased blood levels of procalcitonin, a peptide hormone produced by the thyroid gland, lungs and intestine, are associated with more severe forms of COVID-19. Another meta-analysis of nine studies, involving 1,779 patients, showed low platelet counts to be associated with an increased risk of severe disease and mortality in COVID-19 patients. A systematic literature review of 34 articles describing clinical biomarkers for COVID-19 found reduced lymphocyte and platelet levels in severe COVID-19 patients.

Meta-analyses can guide physicians on what to test once a patient has been admitted to a hospital or other monitored environments. The work published in Cell and Nature Machine Intelligence would permit an earlier and more effective triage of patients. Though these results are encouraging, independent researchers still need to verify them with larger groups of patients. More stringent quantification of the biomarkers and knowing how early in infection this stratification is possible, and will be necessary before these findings can guide clinical decisions. Technology-driven discoveries such as these call for equally robust trials for independent validation.

“While this is an important step forward, such studies should be carried out on many populations across the world if this approach is employed for prognosis,” L.S. Shashidhara, dean of research and professor of biology at Ashoka University, Sonipat, said.

Such technology platforms and expertise are available in several Indian institutions, and now there are patients as well. We must bring these resources together and deploy them thoughtfully to save lives.

Dr Shahid Jameel is a former group leader of virology at the International Centre for Genetic Engineering and Biotechnology, New Delhi. He is currently CEO, DBT/Wellcome Trust India Alliance.


  1. A type of white blood cell

Scroll To Top