How primary care can implement patient-centered AI policy
Patients are aware of the risks of AI use, so primary care offices need to implement a patient-centered AI policy enabling disclosure and consent.
As AI adoption continues to grow throughout the healthcare sector, primary care specialties will need to plan for patients who are both excited and fearful of how the technology affects their care.
A new report from the Commonwealth Fund, completed in partnership with the Public Policy Lab, aims to provide some guidance for primary care offices. The report is based on qualitative interviews with patients in New York City and rural West Virginia. It finds that patients have complicated feelings about AI.
"These tools may aim to ensure primary care is coordinated, continuous, comprehensive, accessible, and patient-centered," the report authors said. "However, design and implementation of health technology tend to be top-down, without consideration for patients' perspectives. While the goal may be to improve the care experience, without their input, patients may be distrustful or hesitant to use these tools, and the tools may create inequities."
Notably, patients see the potential for AI to improve their care, recognizing that it could deepen their understanding of their own health and care plans.
But they also have their concerns, like how AI implementation might ignore the patient voice and create more friction for patients already bogged down by the complexities of coordinating their own care.
Primary care providers will need to create detailed AI implementation strategies to ensure the technology truly delivers on its promises.
Patients see the potential -- and inevitability -- of AI
Patients' AI optimism might be largely driven by their understanding that AI is an inevitability, regardless of industry. But that didn't mean they didn't see the value AI could add to their primary care, the report's authors remarked.
For instance, many patients expressed interest in sharing their medical data to help develop more sophisticated AI models, as long as their data is deidentified. Respondents recognized that their data could help boost not just their own care but the care of others, saying their contributions could be a boon for the greater good.
Additionally, patients lauded AI's potential to deepen their understanding of their own care. AI has proven effective in providing plain-language summaries of lab results or clinical notes. Patients also said AI has been useful when they can't obtain or afford a medical appointment, or when they want a second opinion.
AI has been especially useful for people who've previously had negative experiences with a doctor, such as being judged or rushed during appointments. In those cases, AI gives patients another place to go for medical advice.
Patients worry about AI's potential consequences
Patients acknowledged the possibility that AI could cause cognitive deskilling in their healthcare providers, but that's not all.
They are also jaded from previous technological innovations, fearing that AI could create more friction where it is intended to reduce it. This could be especially apparent when AI is inaccurate or hallucinates.
Finally, patients are worried that AI use in primary care could deepen existing health disparities. This would be true of folks with unreliable internet, older devices or lower digital health literacy.
Despite fears, there are certain safeguards that could compel patients to use and accept AI in their primary care.
First, and perhaps most importantly, patients will need assurances that AI will not replace their doctor. They also want disclosure and consent processes for AI use. Patients want this information at multiple touchpoints so they can assess the AI disclosure and ask questions.
Patients also said they want to see AI enable accountability and accuracy in their care. For example, AI can help them translate complex medical jargon and create their own patient-centered "paper trail," the report authors said. This could help patients work with their primary care providers to verify information, dispute inaccuracies, prepare questions ahead of appointments and otherwise advocate for their health needs.
To deliver on patient needs, primary care offices will need to develop detailed AI implementation plans.
How primary care can deploy patient-centered AI
The report authors grouped their recommendations for patient-centered AI development into three key buckets: transparency and consent, accountability and equitable patient benefit.
Transparency and consent
Clinics should explain AI use and consent via the patient portal before visits and with paper handouts upon visit. Clinics might consider when disclosure versus consent is necessary and determine key areas for opt-out, such as ambient scribing, patient message or clinical decision support. To that end, organizations must maintain patient engagement strategies that do not involve AI.
Accountability
Patients want to know that their PCP has the final say in their care. This requires health systems to train their providers in responsible AI use, including using it as a supplement to their professional expertise. Additionally, primary care organizations need to stress to patients that healthcare professionals have primary oversight over their care.
Equitable patient benefit
Notably, patients feared that AI would not center patient needs and could deepen the digital divide. Healthcare organizations should start addressing this by designing AI use cases that focus on patients, such as better comprehension of medical information. This will require organizations to also measure KPIs that center the patient, including patient satisfaction with technology.
Primary care clinics should also do a health equity audit on potential tools, looking for compatibility with older devices, low bandwidth and different digital health literacy levels.
Patient respondents were right -- AI is appearing to be an inevitability in healthcare. But that doesn't mean primary care providers don't need to be strategic about how they deploy the technology.
"For health system leaders, successful adoption of AI tools in primary care will depend on their design and whether implementation is done with patients," the report authors concluded. "Primary care clinicians can help ensure that these tools are introduced clearly, used responsibly, and assessed in light of patients' needs and concerns."
Sara Heath is an executive editor at Xtelligent Healthcare Media, where she covers patient engagement, healthcare policy and health IT.
