Leading health care systems, such as Duke, “have already done a great deal of hard work,” Dr. Sendak said in a LinkedIn post. “They’ve invested in building models, tested them in real clinical environments, iterated based on what didn’t work, and proven what does (at least within their environment). That innovation shouldn’t stay locked within their walls.”
To achieve his business model, Dr. Sendak built Vega Health based on a four-tiered approach. (This approach is detailed in a four-part series of LinkedIn posts published by Dr. Sendak, all of which can be accessed at www.linkedin.com/pulse/healthcare-ai-distribution-paradox-how-solve-vega-health-zbshe/.)
First, the company provides health systems with the infrastructure on which health care data can be organized in a highly structured way. It creates the “pipelines that curate the data needed, maintains data mappings to preserve elements over time, and facilitates the rapid implementation and monitoring of new AI solutions,” according to the first LinkedIn post.
Second, Vega helps clients prioritize their strategic goals first and then choose AI solutions on the marketplace accordingly based on such factors as their client population and the technologies they currently use, rather than committing to a solution before knowing how well it will meet their needs. “This is how AI procurement should work,” says Dr. Sendak. “Start with your strategic needs, access proven models, validate them in your environment, and implement complete solutions that demonstrate value.”
The third tier involves Vega monitoring selected solutions in real time and measuring related outcomes at the client health care institutions, providing objective evidence regarding what works in each unique environment. This can be of significant value to rural hospitals, which don’t have the staffing required to evaluate and monitor AI technologies, Dr. Sendak notes.
Vega’s comprehensive monitoring addresses several questions: Is the model performing as intended and according to the established parameters? Are the solution’s end users actually using it, and is it changing how they deliver care? Is the solution delivering real-world outcomes, such as fewer patients with sepsis or fewer denials of prior authorization for drugs? Are the outcomes providing value to patients, clinicians, and the health system?
Lastly, Vega becomes the innovators’ distribution partner, helping to commercialize their technologies. “Our distribution model creates pathways for continued innovation while scaling what works,” Dr. Sendak says.
Not only must all models on the Vega marketplace have been implemented in real-world clinical settings, with demonstrated results, but evidence of their success must be validated through publication in a peer-reviewed academic journal or as a preprint article undergoing peer review. Furthermore, the developer must complete the Model Facts Label developed by the Health AI Partnership. “This documentation allows for transparency about the model’s development, validation, and intended use before health systems make purchasing decisions,” Dr. Sendak says.
Efforts are underway for Vega’s first customer to deploy a solution created at Duke for emergency department triage and addressing adult deterioration, which was offered in Vega’s initial 12-product launch.
“Clinical laboratories are a huge contributor to the data that’s generated in clinical care,” says Dr. Sendak. “By overlaying algorithms to already-analyzed lab samples, we can provide additional information and value for the existing data.” At Duke, Dr. Sendak helped develop AI tools to predict sepsis risk and improve utilization of lab tests using algorithms that improve outcomes—principles he uses at Vega.
Dr. Sendak is currently meeting with various academic research institutions to identify previously developed solutions that may be added to the Vega Health marketplace and create new solutions in conjunction with those entities.
“We have a lot of experience doing what we do separately, but these elements have never really been vertically integrated,” says Dr. Sendak. “We believe Vega Health’s model will bring enormous potential to rural hospitals and will expand health care today, in part by centralizing some of the functions and integrating them. ”
—Brenda Lange