Israeli-founded medical AI company Aidoc has partnered with 12 major U.S. health systems to create a consortium aimed at testing whether artificial intelligence can help hospitals diagnose patients faster and more safely as demand for diagnostic services outpaces the available physician workforce.
The Diagnostic AI Consortium includes Advocate Health, Cedars-Sinai Health System, Hartford HealthCare, Houston Methodist, Mercy, Mount Sinai Health System, Northwell Health, Northwestern Medicine, Sutter Health, University of Florida Health, University Hospitals of Cleveland and WellSpan Health.
(Photo: Avishag Shaar Yashuv)
Together, the participating systems treat nearly 20 million patients annually.
The consortium plans to develop and test AI-assisted diagnostic workflows across participating hospitals, measure their impact on safety, quality and speed of diagnosis, and establish implementation and governance practices that could eventually be used by health systems outside the group.
The initiative comes as hospitals face growing pressure in areas such as radiology, where demand for imaging has been rising while workforce shortages have worsened.
According to data cited by Aidoc from the Harvey L. Neiman Health Policy Institute, turnaround times for outpatient imaging interpretation more than doubled between 2014 and 2023. The institute has also reported that radiologists have been leaving the profession at a higher rate since 2020 and projected that shortages could persist for decades without intervention.
Aidoc and the participating hospitals argue that AI could help address some of those bottlenecks by identifying urgent cases, prioritizing worklists and bringing potentially critical findings to physicians' attention sooner.
Unlike general-purpose generative AI, diagnostic systems operate in a highly regulated clinical environment and may be required to analyze signals from medical imaging, pathology, laboratory data and patient records.
That creates additional demands around clinical validation, regulatory clearance and ongoing monitoring for problems such as performance drift or bias across different patient populations, equipment and hospital sites.
“For a long time, the industry treated speed and safety as opposing forces in AI: Silicon Valley's ‘move fast and break things’ against medicine's ‘first, do no harm,’” said Aidoc CEO and co-founder Elad Walach.
“This Consortium is built on the belief that it can be done the right way, guided by two principles: iteratively and together,” he said.
Under the initiative, the health systems will work together to design AI-enabled workflows intended to prioritize cases, accelerate interpretation and deliver important results more quickly. They will then compare results across member institutions and use the findings to develop broader governance and implementation practices.
Jeffrey A. Flaks, president and CEO of Hartford HealthCare, said the collaboration would give hospitals a role in determining how diagnostic AI is introduced into clinical practice.
“Artificial intelligence has tremendous potential to enhance the quality, safety, and efficiency of care by equipping clinicians with more timely, meaningful insights that support better decision-making,” he said.
The consortium reflects a broader shift from hospitals experimenting with individual AI tools toward attempting to deploy such systems across multiple departments and workflows.
The long-term goal described by the group is to shorten the interval between a scan and diagnosis, ensure that critically ill patients are prioritized correctly and reduce diagnostic processes that can currently take days to complete.
Aidoc will provide much of the technical infrastructure for the project through CARE, its clinical AI foundation model, and aiOS, its enterprise AI operating system. The company says the platform is used to deploy and monitor AI applications inside hospitals.
Aidoc says its technology is deployed in nearly 2,000 hospitals and supports analysis of about 60 million patient cases annually. The company says its systems have analyzed more than 150 million patient cases overall.
The consortium expects to publish its first results in 2027.