Big Data’s Next Challenge: Heart Failure

Using sensors and predictive analytics, data scientists are tackling one of America’s biggest killers.

Big Data’s Next Challenge: Heart Failure
[Suregery team: Kotin via Shutterstock]

Nearly 6 million Americans are afflicted with congestive heart failure, which is when the heart doesn’t pump enough blood and oxygen resulting in serious problems in other parts of the body. More than half of people who develop the condition pass away within five years of their diagnosis.


But because congestive heart failure (CHF) has certain characteristics that lends itself to big data analysis, it’s become a moneymaker for tech companies–who are also saving lives in the process.

Heart failure is usually caused by pre-existing conditions, and figuring out what those conditions are can save a lot of lives. But by the time patients are in the hospital, complications related to the disease can make it difficult to parse the cause from effect–at least for humans. Smart medical devices recording thousands of data points per second, and digitized medical records which build network connection maps of all sorts of health care factors, are being deployed to unravel the mystery.

Building Smart Hospitals

Atlanta’s Emory University Hospital is a busy, high-profile medical center whose intensive care unit sees considerable traffic. The hospital has partnered with IBM and Excel Medical Electronics for an ambitious project: Smart medical equipment which records between 1,000-2,000 data points from each patient per second, multiplied by 100 patients.

Dr. Tim Buchman, the hospital’s director of critical care, is a tech-savvy physician who works with vendors like these to test out potentially lifesaving technologies. Excel Medical’s bedside monitors are plugged into an IBM analytics platform which parses the data as it comes in, and–hopefully–finds patterns which predict CHF, sepsis, or pneumonia before they happen.

In an interview with Co.Labs, Buchman compared the experimental analytics system to a GPS for care providers. Although he explained that no analytics system could ever replace “a well-trained critical care nurse,” they can help medical professionals make better decisions in high-stress situations, and anticipate changes to the patient’s health.

“If you speak to a critical care physician and ask how many decisions you make on a patient in a given day, it could easily be 30,” Buchman said. “You multiply that by 20 patients in the ICU and you’ll see that we make 600 decisions daily, all of which are based on situational awareness and a great grasp of information. Currently all of that is based on making sure you didn’t miss the right data elements and remembering what happened five minutes ago. This is the problem that we’re taking on. It is a big data problem, but even more important it’s a data in motion problem.”


One example of the prototype system’s effectiveness can be seen in patients with a common heart disorder called atrial fibrillation. These patients often show no outward symptoms, but the disorder is often associated with congestive heart failures. Clinicians at Emory have, in several cases, used a research-grade analytics system to view real-time digital visualizations of patients’ heart rhythms to spot atrial fibrillation in extremely early stages.

Learning More About Diseases

One of the biggest areas for growth at the Advisory Board, a health care technology and consulting firm that works with most major American hospitals, is congestive heart failure and other conditions with a variety of causes such as diabetes and pneumonia. Because insurers, health care providers, and others have an infinitely cluttered ecosystem where different platforms are frequently incompatible, the Advisory Board can make considerable money helping organizations figure out the best analytics path.

These analytics tools are used for various purposes; one example the company cites is algorithms that stack patients to show their likelihood of being readmitted.

Earlier this year, Fast Company had the opportunity to speak with David Chao, the Advisory Board’s CTO. Chao joined the company through their acquisition of Crimson, a health analytics firm, in 2008. A physics PhD-turned-data scientist, Chao describes his job as helping to “connect the dots and ask questions like what physicians are doing in terms of medication lab tests, X-rays, or imaging that they shouldn’t be doing, or what they should be doing that other physicians don’t do.”

A big part of that role is helping to build analytics platforms that can monitor congestive heart failure. Chao, using a hypothetical example, said that “given the details of each individual patient I work with–their socioeconomic data, where they live, access to transportation, all of this works into ensuring coverage for them. If you’re discharged following congestive heart failure, how do you ensure they get the follow-up care they need? This requires care coordination, which is where big data and predictive analytics comes in. It takes into account what a patient did or didn’t do in the past, the details of their zip code, and that can help predict what a patient needs for care.”

The health care system today has so many different data silos, and messy or incomplete data is such an omnipresent worry, that the health care analytics solutions that Advisory Board and IBM are working on are still in their infancy. If deployed correctly, analytics could save lives and money, and place health resources where they’re most needed. But that’s not going to happen overnight.