CHOC Combines AWS Cloud Computing with Cerner EHR for Data Science

22Aug - by aiuniverse - 0 - In Data Science


August 21, 2019 – Children’s Hospital of Orange County (CHOC) is using the AWS cloud platform with its existing Cerner electronic health record (EHR) to give it the computing power to perform advanced data science research, explained William Feaster, M.D., CHOC’s chief health information officer.

AWS gives CHOC access “to an elastic compute environment where we weren’t limited in any way by memory or numbers of nodes that we could apply to a problem. It is unlimited in terms of applications to libraries, machine learning algorithms, and big data tools,” Feaster told

“The other thing that it provides is much easier pre-processing of data, which speeds up the work of the data scientist,” Feaster said.

“The data is stored in the AWS platform in a very easily retrievable format using FHIR [Fast Healthcare Interoperability Resources] data definitions as well as standard Cerner data definitions for our analysis. It has revolutionized our ability to do data science projects much quicker, using larger data sets and more extensive tool sets,” he added.

CHOC was a partner with AWS in developing the application for the Cerner EHR, Feaster related.

After working with CHOC, Cerner and AWS announced last month a strategic collaboration under which Cerner is naming AWS its preferred cloud provider. This expanded relationship is expected to enable better clinical experiences, increase efficiencies by lowering operational burdens for healthcare organizations, and accelerate artificial intelligence, machine learning, and other innovations.

Cerner said that its clients should expect faster advancements while experiencing a lower operational and financial burden as a result of the AWS partnership. Through its existing collaboration with the Amazon ML Solutions Lab, Cerner has developed new machine learning capabilities for clients.

Cerner’s ongoing work with AWS and use of Amazon SageMaker have allowed researchers to query anonymized patient data to build models and algorithms that led to earlier detection of congestive heart failure.


CHOC currently has several projects underway using the AWS cloud platform and the Cerner EHR.

The two projects that have been finished are readmission prevention and early sepsis detection, Feaster related.

“We’ve had our readmission prevention algorithm in place now for approximately a year … We’re using it to identify patients at higher risk for readmission and our care managers, case management and other staff are focusing more effort on those patients. We’re actually seeing a decline in readmissions,” he said.

“The second project is an early prediction algorithm of severe sepsis. When patients show up in the emergency department, they are triaged by the nurse who gets vital signs and takes the medical history of the patients,” he explained.

CHOC can send an alert based on a predictive algorithm that tells the nurse whether the patient is at very high, high, moderate, or low risk of sepsis during the clinical encounter, either in the emergency department or a subsequent hospitalization.

“We have developed that algorithm with our clinical staff input and at their request so that we can identify more patients earlier who could eventually develop sepsis during their hospitalization,” Feaster said.

CHOC has several more project that are at some phase of development. Some of these projects have taken longer to mature because the data scientists are working with the clinical staff to carefully define all the parameters, he explained.

For example, one of the projects that is underway is a predictor of patient no-shows in the ambulatory clinics. Between 10 percent and 12 percent of scheduled patients don’t show up for a visit. That leads to access problems for them and for other patients, as well as financial costs and poorer patient outcomes.

“Reducing no shows is an important initiative that we’ve put a lot of work into and are getting closer to a model,” he said.

“Another project we’re working on is to identify patients who have rising risk, in other words, those patients who have a certain level of risk for subsequent hospitalization, ED [emergency department] utilization, and clinical deterioration. We are trying to predict those patients who have a higher risk six months out, so that early intervention now might reduce their risk at a later date,” Feaster said.

“We’re working on that predictor to feed it back to our care managers in the ambulatory environment to enable them to intervene earlier for those patients who are flagged as having rising risk,” he noted.

In the future, CHOC will be moving more of its applications to the cloud. “It just seems natural, when we think of a new application, if there’s a way to leverage the cloud for that application, we will consider it and do what’s appropriate,” Feaster concluded.

Facebook Comments