Is AI the key to understanding patient experience data?

Understanding generational differences in patient experience

Not every patient experiences the health system the same way, Tanner said. Using NRC Health's database, the company has been able to outline how younger patients might approach their care differently than older ones.

These differences can fall into three buckets: how they get information, how they leave feedback and their expectations for care.

Younger patients, for example, are more likely to use a handheld device like their smartphone to get information about their care. While 72% of Generation Z (Gen Z) and millennials used smartphones and social media to learn about their health and care, less than 50% of baby boomers did the same.

That trend persists when looking at how the different generations rate their care.

"We're still getting feedback from all the generations, it's just the mechanism in which they're choosing to do so," Tanner said.

For example, around 36% of the silent generation and baby boomers are responding to traditional market research surveys, like CAHPS surveys or other surveys that healthcare organizations field. But as generations get younger, their response rates get way lower, Tanner said.

Only about a fifth of millennials and about 10% of Gen Z respond to traditional market surveys. Instead, they leave online provider reviews, with around 29% of millennials and 40% of Gen Z doing so.

Finally, expectations for the care experience also diverge by generation, Tanner stated. For example, older patients tend to value the competency of their provider more than the overall respect the provider shows them. Conversely, younger patients assume their provider will be competent. Instead, they strongly value a respectful patient-provider relationship.

Younger patients are also swayed by bad experiences across the care experience, Tanner added. If the clinical quality and patient-provider interaction were perfect, younger patients would still be deterred by a terrible billing experience, for example. These generations are also put off by long wait times.

Generational differences guide operational decisions

With these insights, health systems have been able to adopt the strategies and tools they need to provide a good patient experience for every population.

When a healthcare organization knows it will get lower survey response rates after treating more Gen Z and millennial patients, it can augment its outreach strategies. This includes scanning Healthgrades, Google, Yelp and other online provider review websites to glean insights. It can also mean setting up QR codes throughout the clinic to elicit more survey responses from younger patients to avoid non-response bias.

"It's important to avoid making assumptions or misplacing people into those generations," Tanner cautioned. "That's definitely not the intent of analyzing data by generations. I think it's the biggest mistake you can make."

Put simply, healthcare organizations should avoid treating a specific population like a monolith.

That can be accomplished by ensuring the clinic maintains multiple modalities for interacting with patients. For example, clinics can accept cash payments and offer Apple Pay. They might set up online self-scheduling and keep some call center staff employed to book appointments.

"The generational view helps make sure you have broad strokes and channels and capabilities covered," Tanner stated. "But when it comes to interacting with any one person, driving that down to individual preferences is what you need to do."

To that end, hospitals and health systems need to know what each patient wants. That can be accomplished through certain opt-in programs, documenting previous patient encounters and gathering patient experience survey data. Organizations can do a lot by knowing that an individual wants to book and pay online but was unhappy with the communication during their most recent visit.

But perhaps therein lies the problem for many health systems. Between patient feedback on clinic operations and their clinical experiences with providers, there is a nearly insurmountable breadth of information healthcare organizations need to manage and, more importantly, understand.

Creating patient experience profiles

As healthcare organizations continue to prioritize better patient experiences, many are looking to build patient profiles based on experience. This might help clinic staff and providers tailor their services and offerings based on what one individual patient prefers.

Companies like NRC Health have been working to respond to that need by aggregating patient experience and clinical data into patient summaries.

But the company's first attempt at those profiles was complicated, Tanner said. Even distilling that information into a one-page summary was overwhelming for users. After creating tailored summaries based on the user -- different information for front office staff versus the clinicians administering treatment -- the profiles were still too overloaded with patient data.

"AI is really well-suited to this problem," Tanner offered as a solution.

The company built Huey, an AI that sifts through patient information to create contextualized summaries for different end-users. The tool looks through the medical record to find out the patient's goals, expectations for care, clinical data, previous experience data and potential post-discharge risks to guide the end-user.

"We can distill that all down into just a couple of talking points for the physician," Tanner said.

For example, the AI might tell the clinician that the patient has had some negative experiences because they have felt like the provider did not believe their symptoms. The AI would then recommend the provider acknowledge the complexity of the patient's disease and practice more active listening to help the patient feel heard.

It's that last part of the equation that's important, Tanner noted.

"It's not just the data, but an action or a tip," he explained. "It's distilling all of these risks and past experiences and then instead of trying to visualize that, it provides two lines and one thing you can do. Reception in that type of approach has been a really great problem to point AI at."

Of course, the AI can't operate on its own. Organizations need rich patient experience data to help the AI create patient insights and recommendations, which means organizations need to survey across the clinical encounter.

That can be intimidating, Tanner acknowledged, because many organizations lack the staff necessary to respond to negative feedback throughout the encounter. Moving into the future, NRC Health is working to experiment with how AI can support patient experience staff in this pursuit by providing prompting and scripting. Ideally, this can streamline staff workflows.

Sara Heath has been covering news related to patient engagement and health equity since 2015.