Colorectal cancer kills more than 52,000 Americans every year. Early detection is known to reduce mortality, yet more than half of adults skip their recommended screenings. And it’s not just a problem for colon cancer. Better adherence to screening recommendations would save many lives and could even reduce costs and invasiveness of care. The thoughtful deployment of AI can help realize these benefits. AI models can assist healthcare professionals flag and facilitate appointments for individuals who are overdue for screening, ultimately leading to higher screening rates and fewer deaths linked to disease. For instance, a specific program utilizing this method saw colonoscopy screenings surge by over 200%, while colorectal cancer deaths were reduced by 43%.
At Pennsylvania’s Geisinger Health System from 2019 to 2022, an AI model analyzed patient level risk factors including blood test results from patients aged 51-75 who were overdue for colorectal cancer screening. Then, the AI model scored each patient’s risk of getting cancer based on their blood markers, age, and gender. When patients scored above 0.150 on the risk scale, the system flagged them for outreach, and nurse coordinators called them to explain their elevated risk and helped schedule colonoscopy appointments.
The researchers studying the efficacy of this model, compared outcomes for patients who were flagged by the AI versus similar patients who scored just below the 0.150 threshold and didn’t receive calls. The flagged patients were 214% more likely to get colonoscopies within three months and 117% more likely within six months. Most importantly, they were 43% less likely to die within two years.
What makes these results striking is not just the numbers, but how the intervention was structured. The AI model did not operate in isolation. It was deployed as an assist to nurse coordinators. By flagging patients at high risk of colorectal cancer, the technology quickly identified those who needed attention the most. And it was the nurses who made the difference. They informed patients of their individual risks, answered questions, eased concerns, and helped schedule colonoscopies at times that worked for the patient.
By designing medical interventions in this way, we can improve health outcomes and provide healthcare systems a way to plan. When an AI model identifies high-risk patients and triggers targeted outreach, hospitals can use that information to anticipate increased demand and the capacity needed to fulfill that demand. Hospitals can estimate how many nurses and staff are needed to make calls and talk to patients; how many coloscopy slots are needed; and how quickly these patients can come in for screenings.
But this only works if the impact of these interventions is measured carefully. Without knowing how many screenings are done as a result of the use of AI models, hospitals may either overextend their limited resources or leave patients waiting. Either way, in the absence of impact measurement for these models, health systems risk failing to deliver the very care they aim to increase access to.
The lessons from this program extend far beyond colorectal cancer. Other areas of preventive care can benefit from interventions where AI helps healthcare providers understand a patient’s individual risk. By flagging high-risk patients, AI can provide healthcare providers enough information to create customized screening plans, which explain a patient’s risk, outline the appropriate cadence of screenings, and guide them through the steps needed to prevent disease before it develops. When AI is used in this way, it doesn’t and shouldn’t replace human care; it enhances human care, making preventive medicine more targeted, personalized, and effective at saving lives.
This approach goes beyond prediction, it saves lives and offers health systems nationwide a blueprint to improve cancer outcomes while making smarter decisions around resource allocation. The promise of AI in healthcare lies not in replacing doctors or nurses, but in helping them increase access for patients who need care the most. Health systems can turn insights into action, ensuring that preventive care actually happens by combining accurate risk prediction with human outreach and careful planning. Programs like the one at Geisinger demonstrate that when AI is thoughtfully deployed, it can transform regular screenings from a missed opportunity into lifesaving intervention. Applied broadly, these programs can save countless patients across a variety of preventable diseases, and make preventive care accessible, timely, and effective for every patient who needs it.
Author bio:
Professor Carri Chan is the Cain Brothers and Company Professor of Healthcare Management at Columbia Business School and Faculty Director of the Healthcare and Pharmaceutical Management Program. She also leads the school’s AI+Healthcare initiative.
Photo: Supatman, Getty Images
