AI for care of seniors: Who’s minding the recipe—and whose taste buds should hold sway?

It takes a team to develop, finance and implement a new healthcare AI model. As with any group endeavor, goal-focused cohesion can dissolve. This is a real risk when contributors feel there are too many cooks in the proverbial kitchen. 

Concentrating on differences in priorities and decision-making styles within teams working on AI for geriatric healthcare, Johns Hopkins researchers observe the problem and offer solutions in a study published July 29 in JMIR Aging. 

Nancy Schoenborn, MD, MHS, and colleagues ran the research by conducting and analyzing semistructured interviews about tech-enabled care with 49 participants. 

The cohort included 15 older adults (potential patients) and their care partners, 15 clinicians, eight health-system or insurance leaders, five investors and six technology developers. 

The researchers found that, unsurprisingly, the developers and investors clashed with end-users—patients, care partners and clinicians alike—mainly over priorities. 

Areas of across-the-board agreement emerged around adopting AI for older patients, their caregivers and their clinicians. These included cost, value and usability. 

But even there, participants emphasized varying aspects of each concept:  

• Older adults and care partners prioritized low out-of-pocket costs and physical/sensory accessibility.
   
• Clinicians focused on patient affordability and preventing workflow burnout.
   
• Payers and health-system leaders were interested in overall return on investment, system integration and impact on high-cost health events.
   
• Developers and investors prioritized market size, scalability and profit margins to offset high regulatory and development risks.

Among the fixes the participant field recommended were problem-driven design, greater stakeholder engagement, public-private partnerships and educating older adults about AI.

Whose needs are being served? 

AI is unique in that the same underlying product can be used in multiple contexts, Schoenborn and co-authors remark in their discussion section. 

For example, they note, large language models like ChatGPT can be a direct-to-consumer tool or a clinician-facing decision support system like OpenEvidence.

In contrast, prior health technologies such as patient portals have “much more defined and stable” target users, the authors point out. 

“AI’s adaptability means the target user can shift after a product has already been developed and deployed,” Schoenborn et al. comment, adding that such changes have important implications for stakeholder alignment.

“The cost, usability requirements, accountability structures and regulatory pathways may differ substantially depending on the target user,” the authors state. “This flexibility creates both opportunity, allowing products to find their highest-value use case, and risk as it can obscure whose needs the technology was actually designed to serve.”

‘One size fits all’ doesn’t fit 

Schoenborn and colleagues suggest their findings can help inform future R&D of healthcare AI. 

For instance, they write, different elements of cost, usability and value “need to be separately considered and evaluated to meet the needs of diverse stakeholders.”

More:

‘Implementing participant suggestions and other strategies to better align these priorities is essential for ensuring the widespread adoption of patient-centered, impactful AI solutions for older adults.’

The study is posted in full for free.