Convincing C-suite to invest in AI: A new mode of ROI

While it is harder to convince the C-suite that investing in advanced analytics tools makes sense, key pillars in the ROI statement should make that clear.
By Tom Sullivan
08:32 AM
Share

BOSTON – Artificial intelligence and machine learning are going to require new ways of approaching your data for the greatest returns on investment.

“You have to start thinking about using your data different,” Cyft CEO Leonard D’Avolio said at the HIMSS and Healthcare IT News Big Data and Analytics Forum Tuesday. “Amazon does not buy a product that allows them to do population book selling — they’re integrating learning at every step of the way.”

Yet, one of the highest hurdles speakers highlighted is simply getting the funding to start AI work.

That’s because AI, machine learning and predictive analytics rattle the normal mode of return on investment such that many CEOs and CFOs don’t understand the potential, said Leigh Williams, administrator of business systems at the University of Virginia Health System.

“We’re changing the future from what it would have been without this intervention,” Williams said.

That makes it harder to convince the C-suite that investing in advanced analytics tools makes sense. So Williams and David Stewart, senior director of population health and value-based performance at Huron Consulting Group shared tips outlining the upsides in a way non-technical executives can grasp in an ROI statement.

Williams recommended four pieces to include in such a statement: your current capacity to perform the intervention, the effectiveness of the intervention for a specific patient or population, the ability to drive adoption of the predictive solution and patients’ willingness to engage with the intervention. 

The current capacity segment should include an assessment of all the other things you’ll need, such as nurse training, or capital expenses including hardware devices or software.

To describe effectiveness, focus on delivering on promises, and that involves looking to see after an intervention is in place whether it is effective. Understand when you expect to see a change and when to alter the course if you’re not, and commit to a timeframe in which you want to demonstrate improvement. 

When it comes to patients, make clear the power of an individual to shape what’s happening. Take into account what patients actually have to do — taking medications, turning on home monitoring device, etc. — and create a low barrier-to-use to encourage patients, rather than drive them away.

It’s not just about the ROI statement, of course.

Stewart recommended getting the IT and data science teams to interact with clinicians. Take them to the OR or the ED to see firsthand how things actually operate, witness the workflow, and empathize with users’ specific pain points.

“Put what you’re doing today versus what you could do with the tool and prove it,” Williams added. “Don’t just think of this as a one-time investment because if you’re only thinking about the solution in not real-life workflow terms it won’t be able to go.”

And sell the upsides about which areas within the hospital are ripe for significant impact by investing in new AI or machine learning tools.

D’Avolio pointed to reducing unneeded admissions, improving patient satisfaction, and increasing reimbursement as areas ripe for AI.

“We’ve seen other companies in other industries transform from learning, focusing on high priorities, not just focusing on one type of data,” D’Avolio said. “They use all their data, not looking to avoid guardrails but to answer as many questions as they can.” 


 Read our coverage of HIMSS Big Data & Healthcare Analytics Forum in Boston.
⇒ Change management is IT's greatest challenge in delivering needed innovation
⇒ Healthcare is swimming in data, but what to do with it?
⇒ It takes a lot more than analytics to make population health work
⇒ Go figure: The key to big data is actually small data
⇒ Geisinger, Mount Sinai execs offer best piece of advice on succeeding with analytics
⇒ Turning raw data into better health outcomes is no easy task