Meet the Health-Care Company IBM Needed to Make Watson More Insightful
With the addition of a huge medical records database—and the computing infrastructure to analyze it—Watson could help doctors make better use of clinical data.
Large stores of digital health records may contain valuable insights that could improve medical research and patient care.
IBM could be gaining a crucial tool in its effort to turn its Watson artificial-intelligence technology into a valuable source of medical insights.
IBM has said that one of the most valuable applications of Watson, best known for beating human competitors on Jeopardy!, will be its ability to draw practical knowledge from the analysis of massive stores of digital health-care information (see “IBM Aims to Make Medical Expertise a Commodity”). To get one such enormous database, this week IBM announced it is buying a startup called Explorys, which was spun out of the Cleveland Clinic and owns anonymized records pertaining to more than 50 million people. But the company does more than merely hold the records. Explorys CEO Steve McHale says the startup has computing technologies that can mine databases for insights applicable in many areas of health care, from medical research to direct patient care.
Explorys’s technology runs on Hadoop, a data management system that relies on distributed computer clusters and allows for much faster processing and analysis of large, complicated data sets than conventional database systems can handle. The speed was important in the company’s pilot application, a search engine that gave physicians at the Cleveland Clinic a way to find data in medical records. Explorys has since built analytics tools that perform specific functions like identifying risk factors and common gaps in patient care that can lead to bad outcomes.
The technology has also shown it can vastly speed up medical research. Recently, physicians at the MetroHealth system in Cleveland and Explorys employees used the platform to replicate the results of a study on the prevalence of blood clots in a group of patients. The original study required researchers to follow patients for 14 years; by analyzing medical records that had the same amount of data, the new technology delivered the same findings in just three months.
The health records come from a number of health-care organizations that have agreed to help Explorys aggregate data. McHale says identifying information on the patients has been stripped out.
Watson has been put to work analyzing other sets of medical records, helping hospitals look for billing errors and even giving doctors advice in some pilot programs. Now, McHale says, Explorys should “accelerate” Watson’s progress and lead to analytics tools that wouldn’t be possible otherwise.
Keep up with the latest in AI at EmTech MIT.
Discover where tech, business, and culture converge.
September 11-14, 2018
MIT Media Lab