NEVADA, UNITED STATES — A recent guest article published in Healthcare IT Today forewarns the U.S. about its AI healthcare race. Experts wrote that unless the country’s health system addresses its biggest barrier, it won’t be able to achieve its ambitious AI-driven care.
Co-written by Richard Ricciardi, a Professor and Executive Director of The George Washington University’s Center for Health Policy and Media Engagement, and Michael Savas, a fellow at the Center, the article addresses the current state of AI usage in healthcare. It highlights one of the Trump administration’s executive orders titled “Unlocking Cures for Pediatric Cancer with Artificial Intelligence” as the country’s most significant yet drifting initiative as an example.
Well-Intentioned Policy Without Meaningful Interoperability
The executive order directs federal agencies to drive innovation and leverage AI to “prevent and treat childhood diseases.” This signalled a massive step the U.S. took toward embracing technology for the betterment of mankind. With it, faster diagnoses, better treatment developments, and higher odds of survival through AI were supposedly promised.
However, the authors emphasized that unless the country’s healthcare system truly understands how to manage and learn from available data, this promise will remain unfulfilled.
They underlined that the biggest barrier the system has yet to overcome is its disconnected health systems, records, and research silos that have done nothing but lock vast amounts of patient data. This has long stunted innovation, “fueling inefficiency, poorer outcomes, and widening inequities, while inhibiting collaboration across specialties,” according to the authors.
They also warned that these policies, no matter how well-intentioned, will continue to fail in establishing effective operations unless this barrier is addressed.
The Path Forward: Unified Systems and Clear AI Learning
The only way the U.S. will be able to innovate at scale is by nationalizing its health systems with clear AI continuous-learning models. However, centralized data collection has long been considered taboo in the country, as it raises concerns about surveillance and cybersecurity.
Fortunately, Ricciardi and Savas have proposed a resolution for this issue: federal learning.
This system was first proposed by Google researchers in 2016 and allows AI models to be integrated and trained in systems where data already exists: clinics, hospitals, and research institutions, instead of moving them elsewhere.
They believe federal learning is key to reshaping American healthcare and advancing its AI-driven care objectives.
