Sometimes it takes a new innovation to “fix” an earlier innovation.  Or at least to help it come closer to its original promise.

That’s one way to sum up a recent comment at HITConsultant by David Dyke, VP of Product Management, Provider & Life Sciences, for Ciox Health, a health technology company.

At issue for Dyke is the continuing, years-long effort to render medical records interoperable for efficiency, care treatment and research efforts. 

“Generations of medical records are sitting in countless VA hospitals warehouses, and in the databases of every major insurer, on the campuses of research universities and in the offices of every doctor large and small around the country,” he says. “Within those records are medical outcomes for everyone who ever visited a doctor, the longitudinal data of broader populations, the hidden warning signs of future diseases, the untold answers to drug efficacy questions, the indicators and the contraindicators.”

Yet after billions of dollars, the healthcare sector is unable to access most of that data.  In Dyke’s view, however, AI may provide the solution. First, he notes, “(c)ognitive technologies are being used more extensively than ever in health data to guide people through operating complex data processes, to consume the growing volume of health information and to extract actionable insights from the noise. There are numerous successful examples of applied AI in health data that are today delivering highly accurate information treatment and workforce augmentation.”

But as big a step as it would be, Dyke sees “record extraction” as only the first step in applying AI to medical records. Imagine, he says, if AI were combined with blockchain technology. “AI ‘agents’ could one day soon be deployed on a health information blockchain to facilitate the discovery, development and delivery of an unprecedented level of personalized medicine. 

If the blockchain were populated with health data, a swarm of AI agents — built and certified by leaders in the medical community — could search the records of individuals and populations in real-time, reflecting on each record as a single instance in a longitudinal databank. The AI agent could search for health-related warning signs and red flags that a doctor could easily miss in person and could aggregate the trends and insights of giant populations, predicting diseases in individuals, and outbreaks and overlooked cures for populations.”

In Dyke’s view, what all this could lead to, of course, is better care, both on the micro and the macro levels.

“If doctors have access to a patient’s entire medical history at the time of their visits, they become armed with far more meaningful information,” he argues. “If a research university could pull every cancer-related treatment outcome of the last five years with a few simple clicks, they would have far better statistical information at their fingertips as they search for the next generation of cures.”

To be sure, there are miles to go, and similar promise has been imagined before, only to be, at best, greatly deferred. Still, Dyke remains optimistic, noting, “At its core, AI and machine learning may today be used to automate the digitization of medical records, but one day soon, the same tools could be the silver bullet.”