Oasis of Opportunity: Strategies for Managing Healthcare’s Data Deluge

While the digital deluge left most healthcare systems and providers across America swimming in an ocean of raw data, UCHealth perceived an oasis of opportunity. The 12-hospital network spread across Colorado seized on the myriad new forms of data collection and advanced analytics to rethink and reinvent how it provides care to thousands of its patients. An outgrowth of that process was the system’s Virtual Health Center, a centralized command center five miles from UCHealth’s largest medical campus that allows around-the-clock surveillance of individuals both in and out of the hospital while removing much of the clinical burden from the shoulders of overtaxed, overworked physicians.  

More than a cohesive team of physicians, nurses and technologists under one roof, the Virtual Health Center is an instructive roadmap for how healthcare can cut through the noise of today’s bottomless data to catch meaningful signals that flag when a patient is approaching—or actually in—danger. The center marries UCHealth’s electronic health record system with a cloud-based data warehouse where artificial intelligence (AI) and machine learning churn out a reliable, minute-by-minute “fingerprint” of patients who deviate from the norm. The approach has allowed UCHealth to launch, among other things, a surveillance project at one hospital to monitor patients in danger of sepsis—and intervene a critical two hours earlier than in the past. The Virtual Health Center also is among the reasons the College of Healthcare Information Management Executives (CHIME) named UCHealth to its “Most Wired 2019” list of highest-level achievers in healthcare IT. 

“Our hospital system is so complex and chaotic, with alarms going off constantly and data being sent everywhere at all times of the day, that it’s impossible for clinicians to have their finger on the pulse of everything,” acknowledges Steve Hess, chief information officer for UCHealth, headquartered in Aurora, Colo. “But having this kind of eye in the sky helps local caregivers monitor and surveil their patients, look for things that are deteriorating or out of the ordinary, and then work with the local care teams on appropriate intervention. This is where healthcare has to be in the future.”

If medicine has excelled in finding clinical answers to intractable diseases, it has lagged well behind the rest of the economy in digital innovation. Indeed, an article in NEJM Catalyst (online May 4, 2017) reported that while 30 percent of the world’s stored data was generated by the healthcare field, just 180 of the approximately 6,000 data scientists in the U.S. worked in hospitals and health systems. This prompted the authors to suggest that 10 to 20 times more data scientists are urgently needed in healthcare and that data analytics should be established as a core competency across the healthcare realm. 

“When you look outside healthcare, at sectors like finance, entertainment and consumer products that have faced the same problem of exponential growth and too much digital data, they’ve all employed advanced analytics to become more efficient,” says John Rumsfeld, MD, PhD, chief science and quality officer and chief innovation officer for the American College of Cardiology (ACC). “They’ve shown if you use AI, for example, to coax from massive amounts of data precisely the information you need at the right time, you can improve your business by driving greater customer satisfaction and employee engagement. That is what healthcare needs to do now. It needs to undergo a digital transformation.” 

HAVE WE REACHED THE LIMIT?

Whether healthcare is struggling with a data tsunami isn’t in question. Nor is the debate over whether the data deluge is showing signs of easing—no one is forecasting less data for tomorrow’s health providers to absorb. In addition to the endless flow of lab tests, order requests, narrative notes and safety reminders that jam the circuitry of electronic health records, there’s the unrelenting stream of data from insurance claims, value- and cost-based metrics, registries and benchmarks, clinical studies and, more recently, from wearable devices and biomedical sensors that are empowering patients to collect more and more raw information, which physicians are under increasing pressure to review. If that hasn’t created enough angst and pushed the burnout level among professionals struggling for face-time with patients, there’s the data coming from the next digital wave, which is expected to include more imaging (think handheld ultrasound devices that plug into patients’ cell phones) as well as new genomics and biomarker information. 

One way to reframe the issue is to ask if healthcare has reached its data limit. Has it come to a crossroads where any more data is just white noise that contributes little to the field’s real mission of diagnosing and treating disease?

But the more relevant question for many digital and healthcare experts at the start of the 2020s is this: How will we see the signals through all the noise?

For Richard Milani, MD, chief clinical transformation officer for Ochsner Health System, Louisiana's largest healthcare provider, the sight line is quite clear. “What was once thought of as too much data is now seen as extremely valuable because it allows us to do machine learning,” he says. 

A case in point, he offers, is Ochsner’s digital medicine program, which collects reams of data from thousands of patients enrolled in hypertension, diabetes, COPD and other disease-linked satellite programs across its network of 40 hospitals and 100 health and urgent care centers. The data is then filtered through a set of AI-driven algorithms residing in Ochsner’s Epic electronic health record, which sorts, curates and presents the data in dashboard form to a dedicated team charged with evaluating and managing it day today. 

“It gives us two sets of eyes looking at the patient: human and AI,” notes Milani, a cardiologist who has been with Ochsner since 1991. Better yet, he elaborates, the enormous amount of data entering the system bypasses the front-line physician but still allows for a quick medical response if an abnormality is detected. Patients with high blood pressure or type 2 diabetes, for example, use a digital blood pressure cuff or glucometer to automatically upload readings to their MyOchsner account. Their performance is tracked and managed remotely by pharmacists, clinicians and “personal coaches,” enabling 79 percent of individuals with high blood pressure and 81 percent of diabetes patients to reach their goals within six months, according to Ochsner’s website.

The beauty of having AI and machine learning at Ochsner clinicians’ fingertips is the predictive power, Milani says. For example, he and his technology team developed a model that scours the health records of patients to predict which of those admitted to the hospital are likely to rapidly deteriorate and code within four hours. This insight led to the roll-out of a rapid-response team that has significantly reduced the volume of codes on hospital floors. Another AI-driven tool has allowed Ochsner to anticipate which patients are most susceptible to hospital-acquired infections and proactively take preventive steps. The result, according to Ochsner’s website, has been a 50 percent reduction in potentially fatal hospital-acquired infections.