Innovation is expensive, both in terms of budget expenses and operational burden. While it can be easy for those working within the guts of a new tool to see the benefits it could bring to their healthcare organization, these efforts will never have the chance to flourish without support from the ground up — a challenge those who have spent their last few years in healthcare analytics know all to well.

“We had to figure out … the value proposition of analytics,” Dr. Bala Hota, VP, chief analytics officer at Rush University System for Health, said today when describing his group’s early days here at the HIMSS Machine Learning & AI for Healthcare conference in Boston. “This is the choir here, and we all know what [the value] is, but the president, the CEO, the organization hadn’t invested in our team. So that was a gap — we hadn’t made the case for why we mattered, and so we really had to figure out what caregivers really needed and wanted.”

For Hota and his fellow presenters, the most direct way to stake their claim was to target and execute a clear example of how a well-supported data analytics platform could benefit care.

“Build trust by executing. It’s common sense, but if you can deliver value, then people will respond and they will want to go to the place where value is being generated,” he said. “The way I sort of summarized that for the team is ‘Let’s find really great problems to solve.’ If someone can give you a really interesting problem you’re lucky.”

At NorthShore University Health System, this approach took the form of a predictive analytics tool that examined the organization’s standard policy of universal MRSA testing, Chad Konchak, assistant VP of clinical analytics at the system, said. Due to the test’s high rate of false-positives, another layer of data-driven scrutiny had an opportunity to cut back on unnecessary testing and win his grassroots effort broader recognition within the organization.

“It was hard work behind the scenes to build out this information infrastructure, but it was well worth the investment,” he said at the event. “Bottom line, we cut our MRSA tests in half while maintaining our low infection rates, and saved the organization, we estimated, $750,000 a year. We had just started our analytics program, and people started to notice, ‘Hey, there’s something here. Predictive modeling isn’t just kind of a fad, this saves us real money and can help us support our patient outcomes too.’ So this was a huge steroid into the growth and maturity of the team.”

Back at Rush, Hota’s team was able to develop an analytics architecture that could be employed for a range of inputs, KPIs and implementations, which he pitched as a means to streamline the scattered, small-scale analytics projects being conducted across the organization. Among the early challenges his group was able to tackle with this system was a “left without being seen” model that could predict whether patients who were seeking care at the emergency department were at risk of leaving before they received care, thereby preventing lost revenue.

This project won over the support of Rush’s emergency department, Hota said, and was largely powered by the resources already available within his organization.

“We had a big footprint of analysts, actually, that were being ineffectually used," he said. "A lot of the manual work people were doing, you could automate that whole thing. They’re then producing insights. … and the organization liked that because it’s not a net new expense, it’s really an investment in the existing people. That’s an easier case to be made.”

But despite these and other early cost, outcome and care quality demonstrations, both Hota and Konchak said that programs like theirs can only achieve their maximum impact with long-term planning and continued support from clinicians and executives.

In other words, don’t rest on your laurels.

“The reality is we still have a small team, so how do we advance the size of our team and get better? And maintenance of everything we’ve built is another thing we’re now thinking through and planning for,” Hota said. “We’re actively going and growing the team, looking for opportunities to prove our value still, and demonstrate the value proposition of analytics. And maintaining that high level of engagement is another critical factor: you need partners, you need the organization to support what you’re doing, and without that you won’t be successful.”