Skip to content Skip to navigation

Key Datapalooza Focus: Integrating Patient-Generated Data into Clinical Workflow

June 2, 2014
by David Raths
| Reprints
Consensus is that patient-generated data makes more sense in a value-based-care world, not fee-for-service

On the same day that Apple Inc. introduced its HealthKit tracking platform, informaticists gathered in Washington, D.C., to talk about how to integrate patient-generated data into clinical workflows.

During a Health Datapalooza panel discussion, Evolent Health’s Shantanu Nundy, M.D., discussed some of his research into clinician perceptions of receiving diabetic patients’ data. “What we found is that the way you design interventions and make data actionable is critical,” he said. “Just giving clinicians patient-generated data is not going to solve the problem. You have to think through how they will use it in a real-world setting.”

Nundy added that working on integrating patient-generated data makes the most sense in a value-based-care world.  But adding the data in a fee-for-service world doesn't make sense, he added. “You run the risk of overwhelming physicians if it just gets added on to a broken delivery model. It could be counterproductive.”

Jay Nagy, associate principal of corporate strategy for The Advisory Board Co., started his talk with a quote from Dr. Gregory Abowd, a distinguished professor at Georgia Tech. In a 2011 keynote address to the American Medical Informatics, Abowd said, “Within five years, the majority of clinically relevant data...will be collected outside of clinical settings.”

Whether that holds true or not, Nagy said, health systems do need to develop some principles for partnering with patients on data. They should clarify workflow and assign expectations regarding the frequency, transmission, storage and usage of data. “They must ensure that there is perceived value to both parties in sharing data,” Nagy said.

Paul Harris, Ph.D., director of the Office of Research Informatics in the Vanderbilt University School of Medicine, talked about ways Vanderbilt has used its data warehouses to do analyses and feed insights back into the clinical workflow. One example is its PREDICT project involving DNA data to predict drug-gene interactions on the individual level and feed it back to the clinical enterprise and to the patient portal.

Now Vanderbilt is looking at how to integrate patient-reported outcome measurement. “It is a really interesting idea to integrate that into our clinical enterpris,” he said. Vanderbilt would like to collect that type of data from patients and feed it back into the personalized medicine enterprise.

Matthew Bietz, Ph.D., an assistant project scientist in the Irvine Department of Informatics, University of California, announced that the Health Data Exploration project from UC San Diego and UC Irvine has been awarded a $1.9 million grant from Robert Wood Johnson Foundation to create a network to study how to share personal data. "The network will be made up of researchers, consumers, companies, and government entities working in this field to enable data sharing,” Bietz said. The network will explore ways to update policies and procedures for managing risks and privacy as well as techniques for analyzing and interpreting this data.

The avalanche of data from devices and social media needs to be harnessed to help with three types of conversations, said John Mattison, M.D., chief medical information officer at Kaiser Permanente. The first is when a patient wants to talk about their condition with someone else, and that data can provide context for the conversation. Another is between the patient and a personal caregiver team, and the third is between a patient and their professional care team.

“The only way to support those three conversations is to use sophisticated visualization tools,” he said. “I think the onslaught of data is a good thing, but without tools to manage it, it is going to get us into a lot of trouble.”

What health systems need is a way to put the values in context for the individual and a way to escalate the care. How do we know when a change in a reading is noteworthy and how do we route and manage that information?

Finally, David Haddad, executive director of the nonprofit startup Open mHealth, described how his organization is developing an open software architecture to break down the barriers to integration among mobile health solutions. Through a shared set of open application programming interfaces, both open and proprietary software modules, applications and data can be "mixed and matched," and more meaningful insights derived through reusable data processing and visualization modules.

“We are working with clinicians to understand which measures are important to them and then going back to toolmakers and having them expose data in that way,” Haddad said. “We are building an app-agnostic platform so you can prescribe the data you want.”