A group of leading healthcare organizations specializing in cancer care have formed a partnership to advance the space by better leveraging the potential of AI. The Cancer AI Alliance (CAIA) has secured $40 million in cash and resources from big tech backers, allowing it to make major strides in precision medicine.
Members of the alliance are Fred Hutchinson, Johns Hopkins, Dana Farber and Sloan Kettering, who will coordinate the new effort. To be exact, it is the cancer research arm of these organizations.
Fred Hutch President and Director Tom Lynch announced the plan on stage at the Intelligent Applications Summit in Seattle, where the lab is located. Madrona, the VC firm that ran the event, was closely involved in the process as an advisor to Hutch. “We believe this has the potential to make a difference. “This represents an unprecedented ability to agree that collaboration makes progress possible,” Lynch said.
He gave the example of a patient with a rare pediatric cancer who went to one center, but the scientific knowledge to better treat it was stranded in another center, wrapped in proprietary methods and handling protocols. Perhaps within 10 years, that knowledge will be filtered through the scientific literature. But as he points out, children with unresponsive leukemia don’t live that long.
Of course, AI is no miracle worker, and just because it strikes a chord doesn’t mean this problem will be quickly or easily solved by a hypothetical cure-finding model. However, if treatments or research that could help move forward are not visible across these organizations, it slows down the entire field.
The problem is that data sharing between healthcare organizations is not straightforward due to regulations, safety considerations, and inconsistencies between formats and databases. Even if Sloan Kettering’s research to help children with leukemia is at Johns Hopkins, there’s no guarantee it can be shared in a way that’s legally and technically feasible.
The new organization aims to solve this problem through federated learning, a type of secure data collaboration where raw data remains private but can be used for the purpose of training AI and other computing systems.
If research organizations can contribute to shared goals, such as drug discovery or training diagnostic models for existing cancers, while complying with HIPAA and other data controls, we are willing to do so. Building a collaborative system according to this model is CAIA’s goal, but there is still a long way to go, according to Jeff Leek, VP and Chief Data Officer at Fred Hutch.
It’s certainly possible, he explained, but it’s a challenge in terms of technology that can only be accessed once the key players are in place. Preparing these cancer research centers and combining the funding and expertise of Microsoft, AWS, Nvidia, and Deloitte was an essential first step, and not a trivial one. Now, actual shared infrastructure, standards, and specific goals (e.g. pursuing models for specific cancers or treatments) can begin to take shape.
The $40 million is a mix of operating cash, services and intangible assets from the four companies mentioned and will be deployed on an unspecified timeline, except that CAIA expects it to be functional by the end of this year. The initiative must “generate first insights” by the end of 2025.
