Northwell machine-learning tool sees COVID’s future

Their domain: With COVID-19 still raging out there, real-time data has never been more important to Northwell Health staffers.
By GREGORY ZELLER //

A new “predictive tool” developed by Northwell Health can spot COVID-19 outbreaks weeks before they happen.

That’s the word from the New Hyde Park-based healthcare system, which on Wednesday announced a first-of-its-kind digital dashboard that monitors system-wide website traffic on the Northwell.edu domain and its many websites, feeds more than a dozen “indicators” into a machine-learning algorithm and creates a two-week forecast “that has closely mirrored caseload to date,” according to Northwell.

The advance-warning protocol was created by the health system’s Customer Insights Group, in cahoots with Northwell clinicians and IT teams. It came together mostly over the summer, while Northwell was knee-deep in coronavirus patients – the health system has treated some 85,000 to date, with 16,000 hospitalized between early March and Labor Day.

Along with all that emergency medicine comes a wealth of data, generated before, during and after each patient’s COVID-19 experience. Focusing on the “before” traffic, the new digital dashboard mines for patterns in Northwell.edu’s digital chaos, weighing “anonymized” indicators (privacy is assured) such as emergency department wait times and patient symptom searches.

Eric Cruzen: Predictive pulse.

The result: a useful tool for helping clinical teams prepare for coming surges, according to Northwell, which is not only deploying the early-warning system in 19 of its hospitals across Long Island, New York City and Westchester, but plans to share the machine-learning algorithm’s source code with other health systems, helping them spot approaching outbreaks.

The idea is to privately and securely tally patients who might be infected but healthy enough not to trigger conventional alarms, according to Northwell Health Senior Vice President and Chief Marketing and Communications Officer Ramon Soto, who noted “a very strong relationship with website activity [and] in-facility COVID-19 patients two weeks later.”

“You’re feeling sick, start doing the research online, but you aren’t so ill that you need to see a doctor immediately,” Soto said. “This is an important development in the fight against COVID-19 and Northwell wants to share this game-changing tool with the world.”

Dissecting the “public mood” of a particular hospital or region’s web traffic isn’t easy; Northwell.edu has registered more than 20 million pageviews since the start of the pandemic, according to the health system, mostly from patients at two dozen geographically remote hospitals and hundreds of urgent-care centers and outpatient facilities.

But “COVID-19 has really pushed us forward in the usage of real-time data,” according to Eric Cruzen, chief medical informatics officer of Northwell’s Emergency Medicine Service Line, who noted a new emphasis on being “proactive instead of reactive.”

“I worry most about an uptick in viral syndrome presentation in our emergency departments, from fevers to low oxygen levels – we see those numbers in real time,” said Cruzen, also chairman of the Emergency Department at Lenox Health Greenwich Village. “I also pay attention to laboratory testing, and because our health system is so large, we can draw real conclusions from the data.

“This new dashboard is predictive and it is amazing,” the informatics officer added. “We use it every day to keep our finger on the pulse of the community.”