How to prevail over a crisis using data analytics

Just as job numbers affect predicted patterns in the stock market, the predictive model assumes that factors such as the current spread of the disease, how quickly it is happening, how many patients are hospitalized, or the ratio of people hospitalized to those in the ICU or on a ventilator all affect demands on the hospital, he says. “It’s not just how many people have COVID but how sick they are that taxes the system. And this model was able to predict those effects up to a week or two weeks ahead of time.” This capability let data analytics help drive clinical care decisions during the pandemic.

By October 2020, the CART data made it possible for NorthShore to restore 89 percent of its historical surgery volumes a month ahead of schedule and, in addition, to demonstrate that the accelerated recovery period had no negative impact on iatrogenic COVID-19 infection and did not result in increased deep vein thrombosis, pulmonary embolisms, or cerebrovascular accident. This is an example of “how a coordinated and transparent data-driven effort that was built upon a robust laboratory testing capability was essential to the operational response and recovery from the COVID-19 crisis,” write Konchak and coauthors in an article published April 20 in Academic Pathology (doi.org/10.1177/​23742895211010257).

The success of NorthShore’s clinical analytics programs also illustrates that health care is making headway in catching up to data companies like Amazon and Facebook in using data and information not only as an artifact of a business but also as a strategic asset to gain an advantage over the competition, in Konchak’s view.

In reality, actionable data is somewhat unusual, he says. “A lot of people build predictive models that predict outcomes, but the vast majority of U.S. predictive models don’t make it into a decision-making process, meaning they are not actionable. It goes into a research publication, which is great, but it’s not actually leading to improved decision-making.”

A former software engineer early in his career, Konchak was later drawn to health care because “there were so many gaps. Even today we have a bloated, overly expensive system with lower outcomes” than other countries experience. “And just from a technology standpoint, there are a lot of opportunities there.” He has found that conceiving of health care systems as complex manufacturing operations has helped him use his business background to good effect in this particular health care crisis.

“Obviously, all the medical science is so critical to health care, but being able to understand the fundamental economics of health care is also critical,” especially when the subject is a $3 billion hospital corporation like NorthShore. So he thinks about the throughput of patients in a health care system and how to optimize it. “It’s similar to a complex factory where you have inputs and outputs, except the inputs are complicated human bodies. And optimizing the process while being safe and minimizing other downstream outcomes like readmissions—that’s an industrial engineering problem more than it is a medical science problem.”

He hopes the Academic Pathology article will be a blueprint of sorts for other laboratories and hospitals that may want to emulate the data analytics run at NorthShore and to maximize their own data’s usefulness. He compares data to a raw material such as iron ore. “Iron ore is pretty useless sitting in a rock, but taking it through the right supply chain and the right kind of enrichment process, you can turn iron ore into steel and build bridges that get people from one community to another. It’s incredibly valuable,” he says. “Similarly, using the ones and zeros of digital data, once you enrich the data with positive tests, with comorbidities, the locations of patients’ homes, and other information, you start to paint a picture that becomes actionable and advances your ability to deliver better health care.”

One limitation of the consulting analytic tools outlined in the article, Konchak says, is that they are restricted by the populations tested by NorthShore, which has an extensive testing capacity but still tests only a sample of the region’s population. A fuller interoperable, intersystem data infrastructure would dramatically improve the analytic tools, he adds.

That’s not all that’s required, however. “A company can come in and plug in an analytics program, but that’s only one-third of the problem that needs to be solved. Establishing a culture of data is critical for the success of analytics,” Konchak says. “You need to have an analytics leader or an executive leader within your institutions who is championing this to really help drive that data-driven culture.” In addition, health care institutions need to partner with engaged physician leaders. “Physicians are the real decision-makers. Predictive analytics and descriptive analytics like the Data CART are just tools to help them make data-supported decisions.” On this basis he distinguishes analytics from artificial intelligence. “Analytics is more like augmented intelligence.”

How can smaller institutions tap into the insight that data analytics can provide? He advises beginning modestly. “I’m in favor of the way we approached it, by finding small ‘use cases’ where you can show the value of the data. Sometimes people jump too far ahead and try to deploy complicated analytics. And those can become less and less transparent to the decision-makers.”

“Start small. Start simple,” he recommends. “A simple algorithm that indicates interventions for someone who has a lot of prior admissions or a lot of comorbidities is going to be more effective than a fancy predictive model that hasn’t been well thought through.” NorthShore’s clinical analytics program that began with a two-person team was helped in its expansion by partnering with the organization’s research institute. Universities are a resource that smaller institutions can similarly draw upon through partnerships, he says.

But whether through in-house or vendor partnership-based solutions, Konchak believes many small community hospitals can up their game beyond Excel spreadsheets and benefit from better analytics. “You don’t have to have a 25-person team with PhD mathematicians to be successful,” he says. With buy-in from an institution’s executives, and with clinical and business leaders on the front lines who trust the process and use it, “really, anybody can do this.”

Anne Paxton is a writer and attorney in Seattle.