Skip to main content
Skip to footer

by Ryan Hamilton
Published on October 21, 2019

This was originally published on Wired.com on October 8, 2019. 

Picture family members escorting a shaky elderly relative into the hospital emergency department after a fall down the stairs. The patient is complaining of a headache, but a crisis doesn’t look imminent. It’s a Friday, a busy night in the local ER. There are 85 patients already on the radiologist’s work list, and a likely two-hour wait before the patient’s massive—and massively dangerous—pulmonary embolism is discovered on the scans. 

That is, until now. After years of digitizing patient records and leveraging the cloud, the healthcare industry has created a massive and still-growing pool of data. That data, used by analytical tools and increasingly machine learning, can drive everything from streamlining hospital workflows to promoting early detection of cancer or a pulmonary embolism.

“Digitization was really the first step,” says Ryan Hamilton, Senior Vice President of Population Health for the global electronic health record giant Cerner Corp. “The real power is in how you get second-order effects out of that digitization.”

Companies like Cerner, AIdoc, and Arterys are taking advantage of Amazon Web Service’s (AWS) high-speed, high-volume data storage, processing and retrieval in the cloud, and machine learning tools to develop and apply machine learning algorithms—driving positive outcomes for patients and medical staff alike. 

The rise of machine learning algorithms in health care

Hamilton says now that machine learning has proven to be a viable resource for healthcare providers, the next step is scaling the creation of intelligence and its integration back into the workflow at the point of decision-making. Cerner itself is building complex analytical tools that draw on the volumes of secure, anonymized patient data it already has access to: medical diagnosis and treatment outcomes, financial outcomes from claims and coding, billing tools, predictive hospital staffing models, and more.

Take, for example, an ER like the one the embolism patient visited. Facilities across the U.S. struggle with staffing challenges. With one of its machine learning algorithms, Cerner can draw on historical data to predict patient volumes and staff and staff the ER accordingly, days in advance. This proactive algorithm helps ensure that doctors and nurses aren’t stretched thin during their shifts, and that patients are seen more quickly and receive quality care.

Cerner hopes to leverage rapidly advancing machine learning through AWS’s SageMaker to explore additional applications, using its anonymized, HIPAA-compliant records. 

“AWS is giving us access to tools and techniques, whether they’re basic building blocks or complex ecosystems, like SageMaker. Historically, that would have been things we had to invest in and invent on our own,” Hamilton says.

Still, Hamilton says, Cerner won’t be able to build all the algorithms the market needs. The company already works with partners to build machine learning models within the Cerner ecosystem. To truly see impact at scale, however, he sees a need for a broader collaboration.

Read the full article here