AI models from Mount Sinai can predict critical COVID-19 cases

Researchers say the new tools, developed using EHR data from the pandemic's first wave, can forecast short- and medium-term risks for patients over the course of their hospitalizations.
By Mike Miliard
11:31 AM

Researchers at Mount Sinai in New York see promise in new machine learning models they've developed that can assess – within key windows of time – the risk of certain adverse clinical events in some COVID-19 patients.

Research published earlier this month in the Journal of Medical Internet Research describes how the algorithms are enabling better insights into potential risks for a diverse group of COVID-19 patients.

Researchers at Mount Sinai's Icahn School of Medicine and Hasso Plattner Institute for Digital Health gathered electronic health record data from more than 4,000 adult patients admitted to five Mount Sinai Health System hospitals from this spring, during the pandemic's first wave.

Clinicians from the Mount Sinai Covid Informatics Center analyzed characteristics of COVID-19 patients – looking at past medical history, comorbidities, vitals and labs – to help predict the risk of mortality, or critical events such as the need for intubation, within clinically relevant time windows.

By predicting risks for time windows of three, five, seven and 10 days from admission, Mount Sinai researchers say the models offer valuable insights to forecast short and medium-term care decisions for COVID-19 patients over the course of their hospitalizations.

For instance, they note that at the one-week mark – the time period that offered the most accurate prediction of critical events while returning the fewest false positives – conditions such acute kidney injury, fast breathing, high blood sugar and elevated lactate dehydrogenase (indicating tissue damage or disease) were the strongest drivers in predicting critical illness.

Older age, blood level imbalance, and C-reactive protein levels indicating inflammation, were the strongest drivers in predicting mortality.

Some experts have made the case that artificial intelligence had a somewhat disappointing showing in the early days of the pandemic's spread. And it's true that bias in certain algorithms might have an adverse effect on some healthcare disparities.

But AI and machine learning have a big role to play in diagnosis and decision support as the COVID-19 emergency reaches its newest peak. So far, an array of promising models, many pushed out to clinicians via EHR updates, have emerged to help detect the disease and assess risk on a population level.

Mount Sinai, in particular, has been innovating its research into COVID-19 over the eight months since it was inundated with patients during the pandemic's early peak. It's created an AI model to diagnose COVID-19 in patients with otherwise normal lung scans, for instance. And has also pioneered the use of Apple Watch to study COVID-19 stress and burnout among healthcare workers.

"From the initial outburst of COVID-19 in New York City, we saw that COVID-19 presentation and disease course are heterogeneous, and we have built machine learning models using patient data to predict outcomes," said Benjamin Glicksberg, assistant professor of genetics and genomic sciences at the Icahn School of Medicine at Mount Sinai, in a statement.

"Now in the early stages of a second wave, we are much better prepared than before," he said. "We are currently assessing how these models can aid clinical practitioners in managing care of their patients in practice."

Added Dr. Girish Nadkarni, assistant professor of medicine in the nephrology department at the Icahn School: "More importantly, we have created a method that identifies important health markers that drive likelihood estimates for acute care prognosis and can be used by health institutions across the world to improve care decisions, at both the physician and hospital level, and more effectively manage patients with COVID-19."

Twitter: @MikeMiliardHITN
Email the writer:

Healthcare IT News is a HIMSS publication.

Want to get more stories like this one? Get daily news updates from Healthcare IT News.
Your subscription has been saved.
Something went wrong. Please try again.