Making a persuasive business case for bigger AI investment
The genre of "Convincing the C-suite to Invest in X" stories is well-trod ground here at Healthcare IT News. In just the past two years we've written articles describing how hospitals can persuade their bosses to pay more attention to – and pay more money for – data-driven quality improvement, information governance, cybersecurity and smarter analytics for value-based care.
Sometimes it's hard to get those who hold the purse strings to do the right thing – to convince them that it's in the health system's best interest to try a test deployment of a new or perhaps overhyped technology, or that there may be substantial ROI in a project whose value may seem intangible at first.
Artificial intelligence is one area where C-suite skepticism may be high. And, it might be argued, justifiably so. You can't go two mouse-clicks without hearing more about the transformational power of AI, but for many execs the concept may seem hard to grasp, overly-ballyhooed, even science fiction-esque.
Still, it's impossible to argue at this point that smartly targeted AI and machine learning tools, with well-deployed algorithms fueled by clean datasets, can drive big and lasting improvements across healthcare. Just look at a handful of recent stories, spotlighting how AI has enabled hospital transformations in perinatal monitoring, cardiac imaging and oncology.
Use cases coming into focus
So how best to make a persuasive case for AI? How can healthcare CIOs and other IT leaders convince reluctant boards and executive level decision-makers that it's a chance worth taking?
There's still plenty of indecision out there, after all. Gartner, for instance, surveyed the spending landscape recently and found that AI and machine learning are hot technologies that have nonetheless yet to find secure footholds at more than a third of businesses across all industries.
Some 37 percent of those polled for one Gartner survey said their organizations are still shaping and defining their AI strategies. About the same number, 35 percent, reported struggling to identify suitable use cases.
Compounding the challenges is that AI is unlike many other IT projects that might enjoy easier appreciation. It's still newfangled and often unknowable – you've heard, perhaps, of the infamous "black box."
Deployments can be expensive and labor intensive but without immediately apparent ROI. Rolling it out can be complex and require unique skills (47 percent of CIOs told Gartner that they needed to hone news skills for their AI deployments). And once the new tools are in place, whether in clinical or back-office settings, they can demand big changes to existing habits and workflows.
Despite the breathless headlines about artificial intelligence taking over the world, McKinsey offered its own assessment of the state of things. Its verdict? Even though AI investment is undoubtedly on the uptick, tripling between 2013 and 2016, to as much as $39 billion, "don’t believe the hype – not every business is using AI … yet," researchers wrote.
"AI adoption is in its infancy, with just 20 percent of our survey respondents using one or more AI technologies at scale or in a core part of their business, and only half of those using three or more," they explained. "For the moment, this is good news for those companies still experimenting with or piloting AI (41 percent). Our results suggest there’s still time to climb the learning curve and compete using AI."
That freedom of experimentation is only possible in healthcare with the backing and full buy-in from boards and execs, however.
"Survey respondents from firms that have successfully deployed an AI technology at scale tend to rate C-suite support as being nearly twice as high as that at those companies that have not adopted any AI technology. They add that strong support comes not only from the CEO and IT executives but also from all other C-level officers and the board of directors.
Which is why it's important to be able to present a clear-eyed case to those in charge of funding: elucidating the huge value that can be gleaned from well-positioned AI projects, while also being realistic about their hurdles, limitations and potentially slow or hard to define returns.
McKinsey's report is bullish on the potential for AI to transform business processes for the better. Among its advice for organizations looking to take those tentative first steps: "Believe the hype that AI can potentially boost your top and bottom line."
Beyond clinical applications
And those two words – "bottom line" – are perhaps the biggest weapons CIOs and health IT pros have when trying to secure the support of hospital execs in this new era of AI.
In Boston this past month, I moderated a panel at the HIMSS Big Data and Healthcare Analytics Forum, exploring the State of the Industry: Machine Learning & AI in Healthcare.
Sharing the stage with a chief medical officer and a VP of clinical informatics, the conversation, unsurprisingly, hewed toward the ways it can be used to advance decision support, fit into physician workflows, hone predictive analytics and boost population health management.
But as we walked off-stage, the third panelist, Sam Hanna, associate dean of graduate and professional studies and program director in healthcare management at American University, chided me for not asking more questions about non-clinical uses for AI.
It's in those financial and operational areas – human resources, talent management, revenue cycle – where AI and machine learning might be best positioned to more immediately deliver some tangible gains for hospitals and health systems looking to justify their investments, he said.
"We always talk about the patient and the clinician," said Hanna in a subsequent interview. "They are critical components, obviously, of our healthcare ecosystem. But patients and clinicians are supported by a number of professions: administrators, logistics people, IT people, finance people, HR people and many others."
Since that legion of professionals are so critical to the success of healthcare delivery, it would be remiss to not keep the front and center in the AI conversation.
"That larger constituency is a massive component of our healthcare system, which in turn is a massive component of our GDP," said Hanna. "If we're going to be leveraging technology going forward – especially in terms of adaptive intelligence and artificial intelligence – those constituents can benefit from that same technology in a different way."
Consider talent management, for one, said Hanna: "When somebody sends a resume for a job posting, that resume goes into a database and that database is sifted and managed through different algorithms that determine which resumes fit as closely as possible to the job description at hand.
And then that gets "further rationalized, until it ultimately arrives at a specific set that would be reviewed by an HR professional who will ultimately send it to a hiring manager," he said. "We're using algorithms in these databases that sift through various resumes and credentials to ensure we're looking at the right candidates. That's a very simple and a very basic example or use case for AI."
Similarly, on the financial side, "having the right predictive model in place can help an organization understand what the repercussions of their decisions are, and how they could make better decisions," he said.
Even clinical applications of AI might best be "sold" to the C-suite by emphasizing the cost efficiencies they bring, said Hanna.
In imaging and radiology, with X-rays and CT scans and MRI Scans, "the machine, once it's trained, seeing thousands and thousands and thousands of images, can be trained on what to look for," he said.
"That allows the practitioner to focus on more complex or severe cases and have this machine learning be alongside them to help focus on the lower-value items and free up more bandwidth. It would be more efficient, it would cost less, and payers – whether they're public or private – would be very interested in something like this because it would both extend care and reduce the cost of care."
Even better, he said, "once you see the benefit and savings of something like this, then you can apply those efficiencies (cost reductions or profits) back into the system to create additional benefits. If you're investing more in the technology, you're ultimately going to arrive at better outcomes."
Investing in people
One thing to keep in mind when building a business case for AI, however, is that it's not just the technology that requires investment, said Hanna.
"What's important to understand is the skill set,” he said. "These are all great ideas, and some of them are already in action. But they didn't just happen. You need people with the right skill set to do them. Whether you're applying this in the clinical or the non-clinical space, you need to think about what does that workforce look like. What are the skills, personalities, and abilities to not just use existing technologies but envision and implement and guide future technologies.
"It's really important for the future of the workforce to employ people who understand modeling," he said. "And I don't just mean basic statistics, but the ability to work with the tools and technologies that allow them to understand why they're doing what they're doing, and not just how."
Hanna said he understands why some executives might be skeptical of AI. But he also says its potential value can't be ignored.
"Do I think it's hyped up? Yes. Because everybody is talking about it. But a lot of people don't understand exactly what it means. AI in its simplest form is to use data to predict. Knowledge is gathered through data that's collected and modeled.
"A lot of people jump right away to AI as a solution," said Hanna. "But it's not going to be the right solution if we don't understand how we're building the AI capabilities, and understand the building blocks we need to have in place so we can have intelligence – real intelligence. Not AI for the sake of AI, but AI that's actually going to be doing something impactful."
Focus on Artificial Intelligence
In November, we take a deep dive into AI and machine learning.