Amazon Comprehend Medical Aims to Simplify Unstructured EHR Data Mining

Jack Murtha
NOVEMBER 28, 2018
amazon comprehend medical,amazon nlp,amazon unstructured data,amazon ehr

Amazon Comprehend Medical is designed to transform unstructured data into healthcare insights. Image has been altered. Courtesy of Amazon.

Amazon is diving deeper into healthcare with the launch of Amazon Comprehend Medical, a machine-learning service designed to help providers and other healthcare stakeholders extract information from unstructured data in electronic health records (EHRs).

Through the new natural language processing service, Amazon is taking a crack at solving a problem that has long burdened healthcare: how to mine actionable insights from messy, inconsistent information, the most common form of patient data. If the tech giant’s effort proves successful, it could carry big implications for improved patient care, clinical research and any number of data-driven healthcare initiatives.

>> LISTEN: Amazon’s Path of Disruption

“Improving patient care through technology is a passion we share with our healthcare IT and ecosystem customers,” Amazon informatics and machine learning experts Taha Kass-Hout, M.D., M.S., and Matt Wood, Ph.D., said in a blog post announcing the service. “We’re extremely excited about the role that Comprehend Medical can play in supporting that mission.”

The data-mining service, which is eligible to meet health privacy law standards, uses natural language processing to unearth data regarding diagnoses, treatment histories, medication dosages, symptoms and more. Ultimately, Amazon Comprehend Medical is designed to boost clinical decision support, revenue cycles, clinical trials and the security of protected health information.

Amazon said the tool can help everyone from healthcare providers and insurers to researchers, clinical trial organizers and health IT experts.

Unstructured medical data, of course, includes free-form but clinically significant information such as doctors’ notes and prescriptions, audio transcripts and radiology reports. Health IT experts and clinicians have spent many hours detailing the importance of the data — and the painstaking challenges of unlocking it. And many have pointed to natural language processing as a potential game changer.

“Identifying this information today is a manual and time-consuming process, which either requires data entry by high-skilled medical experts, or teams of developers writing custom code and rules to try and extract the information automatically,” Kass-Hout and Wood wrote. “In both cases, this undifferentiated heavy lifting takes material resources away from efforts to improve patient outcomes through technology.”

As such, Amazon Comprehend Medical’s proposition to healthcare stakeholders relies not only on its technological backbone but the promise of a simpler, more cost-efficient path toward that data. All of a sudden, the technical and financial complexities of natural language processing could be lessened, at least to some extent, by the Amazon effect. (The company hasn’t listed a price, though rates are available for its regular Comprehend service.)

Despite facing competition from companies already in the space, Amazon leaders are confident, with Kass-Hout claiming the tech is equal to or better than existing software.

So, how does Amazon Comprehend Medical work? The service enables developers to automatically identify certain medical data without establishing a bunch of custom rules or managing new servers or provisions. Customers don’t need experience in machine learning, either.

“Developers only need to provide unstructured medical text to Comprehend Medical,” the Amazon employees wrote. “The service will ‘read’ the text and then identify and return the medical information contained within it.”

The service also highlights protected health information and does not use customer-supplied data to train its algorithms.

Seattle’s Fred Hutchinson Cancer Research Center has already used Amazon Comprehend Medical to analyze millions of clinical notes, according to the blog post. The technology helped reduce per-document processing time from “hours to seconds.”

The move is yet another step in Amazon’s tech-driven, multifaceted entrance into various corners of healthcare.

Get the best insights in healthcare analytics directly to your inbox.

Related
Amazon Alexa and the Potential of Voice Assistants for Healthcare
A Necessary Threat: Disruption Is the Wake-Up Call Healthcare Needs
With $1B Purchase of PillPack, Amazon Fulfills Its Pharmacy Destiny
 

SHARE THIS SHARE THIS
38
Become a contributor