Long COVID Might Be Diagnosed With Nearly 100 Percent Accuracy: Study

A machine learning algorithm can analyze blood biomarkers to provide a definitive long-COVID diagnosis.

Smit/Shutterstock

A simple blood test may finally provide answers for the millions of people struggling with symptoms of long COVID.

New research published in Nature reveals that distinct biomarkers in patients with long COVID allow a machine learning algorithm to diagnose the complex condition correctly nearly every time.

“This is the first paper of many findings where we’re seeing clear abnormalities between a healthy control population and folks with long COVID,” the study’s principal investigator, David Putrino, who holds a doctorate in neuroscience and is a professor of rehabilitation and human performance at the Icahn School of Medicine at Mount Sinai in New York, told The Epoch Times.

How Does AI Spot Long COVID in Blood Samples?

The findings offer hope of objective confirmation to long-COVID sufferers, who represent about 6 percent of the U.S. population, according to the U.S. Centers for Disease Control and Prevention. Many have faced doubts that their persistent fatigue, brain fog, and pain after COVID-19 infection are real.

Mr. Putrino and his team analyzed about 270 patients from three locations—Mount Sinai Hospital, Mount Sinai Union Square, and Yale School of Medicine—between January 2021 and June 2022.

They divided participants into three groups: those with no prior COVID-19 infection, those who fully recovered, and those with long-COVID symptoms for at least four months after infection.

Researchers had all patients complete questionnaires about symptoms, medical history, and health-related quality of life and give blood samples to identify biomarker differences.

A machine learning algorithm was applied to determine which biomarkers best identified long COVID. The questionnaire data were then fed to the program.

The algorithm differentiated between those with and without long COVID with 96 percent accuracy, detecting unique features in long-COVID patients’ blood. The biggest differences involved abnormal immune cell activity, reactivated dormant herpes viruses such as Epstein-Barr, and reduced cortisol levels.

The findings show promise for objectively identifying the “constellation of immune cells” that can be used as a diagnostic for long COVID, Dr. Thomas Gut, director of the Post-COVID Recovery Center at Northwell Staten Island University Hospital in New York, told The Epoch Times.

“Serological markers could be a huge step toward confirming long COVID versus other complications,” he said.

The data also raise questions about the role of COVID-19 in awakening other viruses, Dr. Gut said.

“This data does suggest that there be a link between these viruses.”

Is It Long COVID or Something Else?

A recent analysis in the British Medical Journal by researchers from the UK, Denmark, and the United States found significant flaws in long-COVID scientific literature. Previous studies may have overestimated prevalence because of broad definitions and lack of control groups, potentially feeding public anxiety and misdiagnosis.

“There is a need for better case definitions and more stringent [long-COVID] criteria,” the authors wrote.

Mr. Putrino’s findings offer clinicians guidance for confirming long COVID.

“We can look at hormonal health. We can run a detailed hormone battery. We can run a detailed immune battery. And we can look for evidence of co-infection,” he said, noting that the study also enables doctors to find the exact problem and treat it with medications.

It’s Not in Your Head: Long COVID Is a Real Disease

This new research could offer a way to distinguish long COVID from psychological distress caused by COVID-19 itself. Mr. Putrino emphasized that the key takeaway is that long COVID isn’t a functional disorder without physical signs.

“We’re not looking at a psychosomatic condition,” he said. “We’re seeing very, very clear evidence that this is an infection-associated complex chronic illness, and this study proves it.”

A study published in February in the Journal of Psychosomatic Research found “substantial” evidence suggesting a role for psychological mechanisms in long COVID. The researchers pointed out that it’s well-established that psychological distress isn’t only a symptom but also a risk factor of long COVID, noting that higher levels of personal resilience were associated with lower severity of the condition.

What Comes Next?

The eventual design of a blood test for diagnosing long COVID will most likely necessitate the use of machine learning software and further research, Mr. Putrino said.

The test wouldn’t offer a straightforward yes or no answer because of the diverse symptoms experienced by different patients, he said. Instead, the test would involve analyzing various factors and symptoms, creating a nuanced approach to accurately identify long-COVID cases based on the amalgamation of these findings.