When nurses at Geisinger Health System called patients to remind them about colon cancer screenings, they had a secret weapon: artificial intelligence. A new study shows that tool helped save lives.

The research, published in the journal Manufacturing & Service Operations Management, found that patients who received AI-guided reminders were significantly more likely to get screened and less likely to die from colorectal cancer within two years.

Colorectal cancer is the second-leading cause of cancer deaths in the United States. Yet many people who should get screened never do. Early detection makes a huge difference — the disease is much easier to treat when caught early.

Geisinger, a health system based in Pennsylvania, wanted to fix that. Starting in 2019, they built a computer program that looked at patient data — including blood test results, age, and sex — to find people who were overdue for a colonoscopy. Nurses then called those patients, explained why screening mattered, and helped them book appointments.

The results were striking. Patients targeted by the program were 6 percent more likely to complete a colonoscopy within three months compared to similar patients who did not receive a call. Within six months, that number grew to nearly 7 percent. Most importantly, the program was linked to a 6.2 percent reduction in deaths within two years — a 43 percent decrease compared to patients who did not get outreach.

"Colorectal cancer is one of the most preventable causes of cancer death if we can reach patients in time," said Dr. David Vawdrey of Geisinger, one of the study's authors. "This study shows that thoughtfully applied AI can do exactly that — identify individuals who might otherwise fall through the cracks and connect them to life-saving screening."

Researchers said the study provides a blueprint for other hospitals. It suggests that AI does not have to replace doctors to be useful. Sometimes the biggest benefit is simply helping patients get care they would otherwise miss. The authors note that hospitals interested in similar programs should consider factors like screening capacity and how to communicate with patients effectively.

The findings add to growing evidence that artificial intelligence can improve health outcomes, not just by diagnosing disease, but by helping health systems reach the right patients at the right time.