MD Anderson's chief innovation officer, Dan Shoenthal, says providers aren't behind on AI; institutions are at much the same stage. He urges health systems to share lessons instead of racing to adopt tools, and says patient expectations need more attention.
MD Anderson uses a two-stage review process: business and technology teams first identify where AI could have the most impact, then a cross-functional group reviews requests weekly for risk. The group includes technology, cybersecurity, business, clinical, ethics, legal, contracting and compliance representatives. Shoenthal said experimentation is also necessary because benefits may take time to measure, and staff time freed by tools can go toward other tasks rather than translate into immediate savings.
Shoenthal said MD Anderson does not allow patient data to train models and does not transfer direct patient data to companies in a way that lets them train on it. The center is deploying some earlier Epic features and exploring Hippocratic AI for patient-facing agents. He said such agents could increase patient contact, not replace staff, and would need operational monitoring.
