Artificial Intelligence in Healthcare

How Providers View Artificial Intelligence in Healthcare

A Historical Look

In the general consumer market, virtual reality (VR) is experiencing something of a resurgence. I say resurgence because, as some may not be aware, the consumer-tech industry actually tried VR back in the late ‘80s and early ‘90s. It didn’t go well. The technology for virtual and augmented reality remained bulky, buggy, and expensive, and the graphics experience didn’t live up to the hype. Several videogame tech giants found that when the “smoke-to-substance” ratio of their products is out of balance, they won’t get very far with consumers.

Entertainment wasn’t the only cutting-edge industry with eyes bigger than their stomach. Healthcare IT spent a lot of time in the ‘70s trying to innovate by creating artificial intelligence (AI) in the form of IT systems to aid and ultimately replace doctors in diagnostics.

Healthcare IT companies found that, while the concept for AI in medicine made sense, the market wasn’t ready. Plagued by a lack of trust from doctors and low adoption rates, AI slipped quietly into the background of healthcare IT hype. That is, until 2018, when I couldn’t walk pass three booths on the HIMSS showroom floor without seeing something related to AI. Interestingly, it seems that the AI of today has undergone something of a rebranding.

Instead of touting AI’s ability to replace human diagnostics altogether, the Center for Connected Medicine (CCM) reported that 46% of their respondents use AI in the form of clinical decision support. Population health and disease management came in as the second and third most popular areas of AI implementation. For the time being, it appears that providers have accepted a “man and machine” instead of “man vs. machine” approach to AI.

The Provider Perspective

Unsurprisingly, the Hype Cycle and marketing buzz surrounding AI have outpaced providers in terms of what health systems actually focus on. KLAS recently released a report on the efforts of CommonWell and Carequality to boost interoperability in the industry. I spoke to one provider about using these platforms, who explained that “connecting still feels a little like a bridge to nowhere.” There’s still plenty of work to do in the fundamentals of digital health.

 

Interoperability Chart

 

Indeed, it seems that the provider community feels the need to address several other pressing areas before they giving this second wave of healthcare AI their full attention. The aforementioned report from the CCM indicated that EHR optimization, integration, communication solutions, and cybersecurity all took precedence over AI implementation for health system leaders.

My own recent experiences with providers confirms this. KLAS’ launching of the Arch Collaborative—a provider-led effort centered around the EHR—has spread to more than 100 health systems and encompasses some of the largest providers the US. As the EHR-buying frenzy of the last decade subsides, it makes sense that providers will now look at these expensive systems and ask, “How do we get the most out of this investment?”

This lack of overwhelming interest in AI presents not a problem but a huge opportunity for AI vendors. The infamous AI winter, or long period of cooled sentiment toward AI, came about because the vendors who brought solutions to the table had vastly overstated their impact.

The relatively low demand from providers today means that vendors can (and should) identify a few forward-thinking health systems that are passionate about AI who can act as partners. Together, these organizations can incubate and develop more fully mature AI product offerings. Since nothing cuts through market hype like real-world success stories, this approach can only boost AI’s victories in the long run.

Another key difference between the AI of yesteryear and today’s AI tools lies in big data. As healthcare goes digital, the amount of information our systems spew out grows exponentially. We have already seen big data revolutionize the markets of other industries. For example, the Security Exchange Commission now uses big data analytics to monitor financial market activity for illegal trading. Perhaps the strongest use case for AI in healthcare will come as providers look to the mountains of information piling up inside their EHR and realize, “It’ll be much easier if we let the computers sort this out.”

​Conclusion

Over the next few years, provider focus will naturally shift away from contemporary problems like interoperability and EHR optimization as they are solved. The AI vendors who have spent their time on development (rather than on generating hype for a half-baked product) will have not only deep wells of big data to analyze, but also use cases and best practices under their belts.

This will prove vital for vendors as their customers increasingly look to them as partners instead of simply tool salesmen. That level of customer engagement means vendors will need deep, strategic vision with the knowledge and experience to back it up. Providers are beginning to expect this. As one provider explained, “After all, we’re not buying from a vendor; we’re marrying them.”

 

KLAS, Triple Tree, L.E.K. Consulting and the Marwood Group are hosting the 2018 Digital Health Investment Symposium (DHIS18) August 14th-15th.

DHIS18 will bring together executives from the provider, payer, vendor, and investor communities to discuss topics like artificial intelligence. Learn more.