Amazon technology deciphers text in electronic health record

Amazon Web Services unveiled a new natural language processing service that it says can pull keywords and important messages from text, known as unstructured medical data, providing fuel for population health and clinical trials, the company announced Tuesday.

The service, Comprehend Medical, is designed to home in on diagnoses, medication dosage, symptoms and other information that's often hidden in unstructured text, which, according to Optum, makes up about 80% of information in health records.

Once Comprehend Medical identifies the relevant data, either directly from the EHR or from extracted data, developers can apply analytics for clinical decision support, revenue cycle management and other tasks.

Healthcare organizations might also use those insights derived from clinical records for building clinical trials more quickly and accurately, using automation to improve upon a process that traditionally has required researchers to go over unstructured clinical data manually.

At the Fred Hutchinson Cancer Research Center in Seattle, an early customer of the product, researchers are using Comprehend Medical to index clinical records for research databases, making it easier to find patients for trials and to figure out whether trials are feasible in the first place. "This is a terrific example of a bottleneck that is well-addressed by the judicious application of technology," said Fred Hutchinson Chief Information Officer Matthew Trunnell.

Eventually, Fred Hutchinson will apply Comprehend Medical to all of its patient notes.

The service, which can run on any EHR's data, is HIPAA-eligible, and it doesn't require Amazon to store any information on the back end, and Amazon says it doesn't see any of the patient data.

"We're taking the process down to seconds," said Dr. Matt Wood, general manager of deep learning and AI at AWS. "Reducing that time has a meaningful benefit to providers and insurers. They can spend their research time on more valuable tasks."

The service is an extension of Amazon Comprehend, the company's broader natural language processing service, announced at the 2017 AWS conference.

"This is really where a lot of systems are excited about this opportunity," said Dr. Taha Kass-Hout, a senior leader at Amazon focused on healthcare and artificial intelligence-related initiatives, "since it can enhance their understanding of the health of individuals and populations."