Predictive Analytics is transforming health care

by Karyn Hede, freelance writer in Chapel Hill, N.C.
Published on Sep 19, 2016

Health systems now require enterprise data warehouses to create a foundation for future analytics. Analytics is shifting from explaining what happened in the past to predicting what will happen in the future, allowing proactive interventions. Effective use of predictive analytics requires input from multiple data sources, such as EHRs, lab tests and insurance claims.


Hospitals are on the cusp of entering the era of “Moneyball” medicine. Just ask Eric Topol, M.D., director of San Diego-based Scripps Translational Science Institute. He recently hired Paul DePodesta, a data analytics guru who transformed the Oakland Athletics baseball team and now has his algorithms set on doing the same for health care. And he is not alone. More than 200 data analytics companies are vying for the attention of health care organizations, which are sitting on an untapped trove of data.

With the near universal adoption of the EHR, medium and large hospitals and health systems have begun to recognize something that consumer retailers have relied on for more than a decade: With the right analytics, data can predict the future and help organizations get out in front. In case management, predictive analytic systems are being used, for instance, to understand which patients are vulnerable for progression-of-care barriers, at higher risk for hospital readmission, or to reduce hospital stays after some elective procedures.

Some health care organizations have opted to build their own data analysis systems to suit their needs, while others have found industry partners.  Analytics-provider options range from huge general use vendors such as SAS, IBM and Oracle to niche players developing specialized tools for the health care market. For organizations looking to implement analytics, those who have already made the plunge suggest starting by taking stock of your organization’s current state.

'The first thing you need to know is what is happening in your hospital," says Rishi Sikka, MD, Senior VP of clinical information for Advocate Health in Illinois. Do you know who is being readmitted today?  Do you know who is visiting the ED? Do you know who is on certain drug categories or who has speciic socioeconomic factors that might indicate a need for care coordination or referral to a formal transitional care program.  Everyone want to do all the sexy stuff, but first you got to understand the organizations's current state.  

Operating as a value-based care organization for several years, Advocate has been moving from fee-for-service revenue models to value-based reimbursement, which is driving organizations toward sophisticated analytics systems that can quantify quality measures and track process improvements.  Five years ago Advocate had a big problem with siloed data spread across many EHR systems that did not play well together. With motivation to improve patient care and control costs, leaders chose to invest in a partnership with Cerner, a cloud-based analytics platform that could integrate data from all of the EHRs within Advocate’s existing information technology infrastructure.  “As an industry, when we talk about population health we tend to talk a lot about cost and utilization and those are extremely important … but that’s not something that gets clinicians excited. If you start with the clinical scenario and why is it important to patients … that’s where you really start to capture the hearts and minds of physicians.

Like many hospitals, Advocate was struggling to reduce 30-day hospital readmissions, a key benchmark for Medicare reimbursement. The model developed in partnership with Cerner automated the process, identifying patients deemed at high risk of readmission and embedding the information within the EHR. After a year of use, readmissions from all causes had dropped by 20 percent among the highest-risk patients within the Advocate system when comparing outcomes in the first half of 2013 with the same period in 2014. Cerner has since made the model, now called Project Boost, available to clients nationwide including our client Northern Arizona Healthcare, which reduced its readmission rate by more than 40 percent since implementing the model in mid-2014. 

Bharat Sutariya, M.D., chief medical officer and vice president of population health for Cerner says that his team is implementing a predictive tool to identify patients who could most benefit from care managers for care coordination.  He gives the example of the common hospital practice of waiting until a patient is ready to be discharged to find a wheelchair when it should be possible to anticipate that need well in advance.

Ayasdi, an analytics company, combines machine learning with algorithms that generate geometric patterns in big data. The search structure helps analysts identify data forming meaningful clusters that can be further investigated. In an early use case, one Ayasdi client wanted to understand the factors that influence hospital length of stay after joint replacement surgery.  Searching across the hospital's network, the Ayasdi algorithms identified a cluster of patients with shorter length of stay. Looking more closely, the analyst team identified a group of providers that was using pregabalin, a neuropathic pain reliever, in the acute aftermath of surgery. The patients who received this treatment used less opioid pain reliever and were ambulatory more quickly than other patients.

Some health systems have opted to develop their own enterprise systems from within by hiring data scientists and develping their own predictive analytic systems.  The University of Pennsylvania Health System (HUP) has invested heavily in a centralized data warehouse and development of machine-;learning models that can make forecasts and then push the results back out to HUP's EHR system. Penn's early projects centered on spotting patients at high risk or health failure and sepsis. Until the machine-learning model was implemented, only heart and vascular service line patients were being evaluatd for heart failure, and the system was only capturing about half the patients at risk. Now, all patient deemed at high risk through the machine-learning algorithm are flagged for transitional care and the hospital is piloting a project to connect high-risk patients with home care.   

The value is in the data.  According to HUP's data scientist Michael Draugelis, “With a small team of developers and data scientists, you can build a system. What’s happening in the marketplace right now is there are a lot of vendor solutions out there, but I’m not sure the market is valuing the right thing right now.” Penn plans to scale up its model, named Penn Signals, and publish it as an open source solution for others to use in mid-2016. “What I want to do is make the technology freely available, and I think that’s a first step to lowering the barrier to making these solutions available,” he says.  Draugelis does point out that any institution that wanted to deploy his open-source system also would have to invest in the data scientists who know how to use its capabilities.

For health systems with more modest resources, Seattle-based Tableau offers a data-visualization system that can import and combine data from a myriad of sources and display them in a visually intuitive dashboard. Given its origins as a collaborative visualization project by graduate student Chris Stolte and his adviser, Pat Hanrahan, founding member of the movie animation company Pixar, the software is flashy, with lots of visualization options. Its biggest strength may be its ability to import data from disparate sources and combine them in an intuitive dashboard for data novices. That ease of use may account for its explosive growth. In a 2014 HIMSS survey, Tableau was reported to be the most commonly used data-visualization software.

Rajib Ghosh, chief data and transformation officer at Community Health Center Network in central California, says Tableau helps his organization's eight community health centers manage a population of about  200,000 within their ACO.  For two years, they have beeen using the system to combine financial, clinical and pharmacy data to understand how people were accessing services.  Ghosh says that his organization, has business analysts on staff who can look at the data critcially to solved priority issues.  "Data are tools,"  he says.  "These are the means to an end, they are not an end in itself.  If you don't have a person who can do that and you just have some IT developers, that's not going to cut it." 

 

Portions of this article first appeared in Trustee magazine, April 4, 2016