A 55-year-old man who was initially diagnosed with a malignant but low-grade brain tumor was being monitored at a nationally ranked cancer center, and after seven years of brain scans the tumor board determined that it appeared to be stable. That indicated neuro-radiation or chemotherapy treatments — both of which carry their own risks — did not appear to be needed. But when the medical school doublechecked the images with an artificial intelligence, or AI, application co-developed by professors at Rowan University and the University of Alabama at Birmingham School of Medicine, physicians discovered that the tumor had indeed grown. And a subsequent biopsy showed it had turned deadly, requiring lifesaving radiation and chemotherapy.

AI is helping physicians across New Jersey and elsewhere to raise their level of patient care. But doctors who spoke with NJBIZ say there’s still a place for the human touch. “The tumor board met and, after reviewing our AI-enabled results, a neurosurgeon said that the tumor should have been removed five years ago, and had now advanced too far to be removed by surgery; and if the doctors had waited another three months without administering radiation and chemotherapy, the patient would have likely been paralyzed by the tumor,” said Nidhal Carla Bouaynaya, director of Rowan’s Artificial Intelligence Lab, professor of Electrical & Computer Engineering, and associate dean for Research and Graduate Studies with the Henry M. Rowan College of Engineering.

She also co-founded MRIMath, a company that uses artificial intelligence to improve medical care and patient outcomes. “When an individual has a non-malignant brain tumor, the safest course of action sometimes means doing nothing but monitoring it for years to see if it remains stable. But contouring, or mapping out a brain tumor is complicated and time consuming, and human error may creep in.”

In contrast, the interactive AI-human interface developed by Bouaynaya and MRIMath co-founder Hassan Fathallah-Shaykh, a neuro-oncologist at UAB School of Medicine, can quickly and accurately plot out a tumor’s measurements in three dimensions in seconds and slashes the total time for physician review to three minutes. That’s compared to the 90 minutes it can take a radiation oncologist to just map out a tumor. “This can reduce the risk of misanalysis, since the human element is often the weakest link in tracing tumors,” according to Bouaynaya. “Studies have shown that the results of two radiation oncologists mapping out the same tumor can vary as much as 30%. It’s a huge issue.”

The costs of an AI-enabled approach can be at least partially offset, she added. “Our application, which is currently under FDA consideration, is designed for large and small health care systems. Multiple clinical research organizations have told us that our software is cost-effective, compared to paying a technician’s salary and benefits for this kind of mapping.”

For now, Bouaynaya and her collaborators are focusing on the brain, “But we plan to expand to other organs that may be affected by tumors, including the lungs,” she said. “There’s a lot of ground still to be covered.”

The Implementation of AI and machine learning, or ML, in health care “is a compelling vision that has the potential to lead to significant improvements for achieving the goals of providing real-time, better personalized and population medicine,” according to Zeeshan Ahmed, assistant professor of medicine at Rutgers Robert Wood Johnson Medical School, and the Institute for Health, Health Care Policy and Aging Research. “The rightful use of AI/ML will improve the traditional symptom-driven practice of medicine and allow earlier interventions using predictive diagnostics and tailoring better-personalized treatments.

AI and ML can reinforce the doctor-patient connection. “Technology is contributing to the development of better relationship between medical professionals and their patients,” he noted. “With the right use of the technology, today, medical professionals can provide timely, better, and precise diagnoses and treatments.”

There’s been an “explosion” of AI-related advances in medicine, according to Dr. Devon Klein, chairman of radiology and radiation oncology at Atlantic Health System’s Overlook Medical Center. “It’s a three-pronged development in radiology,” Klein said. “First is the natural-language processing ability of AI, which enables it to comb through massive amounts of radiology reports to quickly analyze and process information and highlight certain trends or keywords. The second prong is the ability to use AI in radiology image acquisition, where it can improve image

quality while reducing the radiation used. Finally, AI can be used in reviewing images to identify critical and other conditions and expedite care, potentially saving more lives.”

The flip side is that AI and other high-tech systems are expensive, often costing “millions of dollars,” according to Sunil Dadlani, Atlantic Health System’s senior vice president, chief information officer. “So we don’t start by selecting the tech system. Instead we use a pragmatic approach, identifying the challenge that we need to solve and then determining the right one to use, the right price and the right time. Then we investigate whether a product has been FDA approved, and whether other hospital systems have used it. Finally, if we do make the purchase, we make sure the vendor has some skin in the game by have them extend some degree of vendor financing.”