Northwell's A.I. investment aims to speed pancreatic cancer care 

Northwell Health is investing $500,000 in an artificial intelligence tool it hopes will speed the diagnosis and treatment of pancreatic cancer and advance research into the often-deadly disease, including why it hits Black patients harder than others.

The tool, called iNav, quickly combs through reams of medical records, looking for indications of a suspicious mass in a patient’s pancreas.

The earlier the disease is detected, "the more likely we'll catch it at a time that it can be removed and potentially cured,” said Dr. Daniel King, assistant professor at the Institute of Cancer Research at Northwell’s Feinstein Institutes for Medical Research and one of the developers of the new tool.

Northwell’s investment comes as artificial intelligence makes headlines for its benefits and potential dangers. In health care, tech experts say A.I. has the power to improve patient care, though it must be monitored to make sure it does not endanger patients’ confidentiality or worsen health disparities and bias.

A.I. tools must be carefully tested, especially since there can be “subtle differences” in the ways diseases manifest in people of different ethnic and racial groups and genders, said Anant Madabhushi, a professor of biomedical engineering at the Emory University School of Medicine in Atlanta.

A.I. “can really be a force for good,” he said. But, he said, "being able to really hone in on how the model is doing across genders, across different populations, and quantifying that and reporting that, I think becomes really critical."

New Hyde Park-based Northwell announced the funding for the new A.I. tool this month at its 2023 Innovation Challenge, modeled after the “Shark Tank” reality show.

Northwell aims to make a multimillion-dollar return on its iNav investment by increasing the number of pancreatic patients who make follow-up appointments at its facilities, and eventually by selling the technology to other health systems, executives said last week. No date has been set to offer the technology for sale.

At Northwell facilities, the new A.I. tool reviews the text of reports about the roughly 10,000 abdominal CT and MRI scans the health care system conducts each week, looking for keywords such as “mass” while disregarding phrases such as “no apparent mass.”  

Even without an A.I. tool, whenever a radiologist finds a suspicious mass, the patient's doctor is notified. However, iNav speeds the process of connecting the patient with medical treatment, King said. When iNav finds red flags, it immediately alerts doctors as well as nurses who help patients navigate the health care system. The tool lets staffers know if a patient needs to talk with a staffer who is fluent in their language, or if they face other challenges, such as living in poverty. It also alerts cancer researchers, who can invite patients to enroll in a clinical trial.

Why early treatment matters

Pancreatic cancer has a five-year survival rate of just 12%, and it is expected to strike more than 64,000 Americans this year, according to the American Cancer Society. Black people are at especially high risk due to factors that include higher rates of cigarette smoking and diabetes, and their cancers are more likely to be diagnosed at an incurably late stage, researchers say. Early treatment can make a difference; the five-year survival rate is 44% for cancers that are limited to the pancreas, but only 3% for those that have spread far from the organ, the ACS reports. 

“If we can pick up lesions early on and navigate those patients to the right doctors, that's important,” said Dr. Richard Barakat, physician-in-chief at the Northwell Health Cancer Institute. “We want to bring cutting-edge treatments into all of the communities that we serve.”

The new tool got its start last year, when King and his colleagues faced frustration in their efforts to enroll more Black patients in a study about racial disparities in pancreatic cancer. Then it occurred to them that A.I. could help. The tool has identified three potential participants for the Northwell and Cold Spring Harbor Laboratory clinical trial, which is still enrolling patients, King said.

The next steps will include using A.I. to also review radiology images and other records that could suggest potential cancer, King said. Already, iNav is flagging suspicious masses within a day of a report being written, and patients are being referred for care within a week, King said. Previously, it took an average of 37 days from the initial scan to the first oncology appointment, he said.

The tool is part of a wave of new A.I. technology tools designed to improve the diagnosis and treatment of cancer and other diseases. Among them is an A.I. tool that can identify people at high risk of pancreatic cancer up to three years before they would otherwise be diagnosed, according to an article published this month in Nature Medicine.

That tool scans medical records for about 2,000 conditions, including abdominal pain and jaundice, said Chris Sander, co-senior investigator for the study and a professor at Harvard Medical School. Still under development, it could be available in roughly three years as “open source” technology that service providers could offer to hospitals, he said.

“We are pushing as scientists to make this happen as soon as possible because it's really useful to catch this aggressive pancreatic cancer early,” he said.

By Maura McDermott

maura.mcdermott@newsday.commauramcdermott

Maura McDermott covers health care and other business news on Long Island.