Harold Robert Meyer and The ADD Resource Center 04/16/2025
Executive Summary
Generative AI presents a promising solution to streamline patient portal communications by generating initial responses for healthcare providers to review and edit before sending to patients. This approach aims to reduce administrative burden while maintaining medical accuracy and personalization. Our analysis explores adoption potential from both provider and patient perspectives, with insights on implementation considerations and potential outcomes.
Why This Matters
The healthcare system faces a critical efficiency crisis. Physicians spend approximately two hours on administrative tasks for every hour of direct patient care, with patient portal messages adding significantly to this burden. Meanwhile, patients experience frustrating delays in receiving responses to their health concerns. AI-assisted communication offers a potential win-win: reducing provider burnout while improving patient experience through faster, more comprehensive responses.
Key Findings
- Provider adoption depends on system integration, reliability, and demonstrable time savings
- Patient acceptance hinges on transparency, maintained personalization, and faster response times
- Implementation success requires attention to tone, accuracy, and proper framing of the technology
- Pilot programs show promising results when human oversight remains central to the process
Note: The National Academy of Medicine’s recent special publication highlights that while generative AI has the potential to profoundly reshape medical practice and improve patient health, its responsible, ethical, and fair adoption demands thorough attention to the risks involved. Achieving this requires active collaboration among clinicians, patients, policymakers, ethicists, and researchers, as well as a broad, cross-sector commitment to maximizing the benefits of generative AI while minimizing its risks—an approach essential for managing the complexities of its integration into health care.
Q: Should doctors use Generative AI to help manage patient portal communications?
A: There’s a compelling case for AI-assisted patient communications, with benefits for both providers and patients when implemented properly
The healthcare administrative burden has reached unsustainable levels. According to the American Medical Association, physicians now spend nearly two hours on paperwork and electronic health record (EHR) management for every hour they spend providing direct patient care. Patient portal messages—ranging from routine medication questions to complex symptom reports—have become yet another administrative bottleneck.
Generative AI offers a promising solution by drafting initial responses based on patient records and medical guidelines. This approach could be particularly effective for handling routine inquiries (e.g., “Should I take this medication with food?” or “When should I schedule my follow-up?”), freeing physicians to focus their limited time on more complex medical decisions.
Q: What would drive physician adoption of this technology?
The key factors influencing physician adoption include:
Reliability and accuracy – AI systems need to consistently produce high-quality drafts (approximately 80% accurate after training on specialized medical datasets) that require minimal editing to be time-efficient.
Seamless integration – The technology must work within existing EHR systems like Epic or Cerner rather than adding another separate tool to physicians’ already complex digital workflow.
Demonstrable time savings – Doctors would be more likely to adopt if the system could reduce their screen time by a meaningful amount (e.g., 30 minutes per day).
Maintenance of medical control – The review-and-edit workflow preserves the physician’s final authority, aligning with how doctors already delegate tasks to support staff while maintaining oversight.
A 2023 survey published in the Journal of Medical Internet Research found that 68% of physicians expressed openness to AI solutions that reduce clerical work, provided these tools don’t compromise care quality. This suggests a receptive audience among healthcare providers, particularly as they’ve begun adopting other AI-assisted tools in areas like radiology diagnostics.
Q: Would patients accept AI-generated responses, even if reviewed by their doctor?
Patient acceptance may be higher than initially expected, particularly if:
Transparency is maintained – Patients know their provider reviews every communication before it’s sent
Response quality improves – Messages are clear, thorough, and directly address their concerns
Response times decrease – Faster turnaround replaces the multi-day waits many currently experience
Trust remains the fundamental issue. While consumers have grown accustomed to AI in customer service contexts, healthcare communications involve higher stakes. A 2024 Pew Research poll revealed that 60% of Americans express discomfort with AI making medical decisions, but this drops to 35% when a doctor maintains oversight of the process.
Patient acceptance would likely be highest when the technology is framed as “your doctor’s team, enhanced by technology” rather than “AI replacing your doctor.” The emphasis should remain on how this approach allows their healthcare provider to deliver more comprehensive information while using their time efficiently.
Q: What implementation challenges might healthcare organizations face?
Successful implementation requires attention to several critical factors:
Tone and empathy – AI must be trained to communicate with appropriate empathy rather than clinical detachment, particularly when discussing sensitive health matters.
Medical accuracy – Systems must avoid “hallucinating” incorrect information, particularly around critical elements like medication dosing or treatment recommendations.
Privacy and security – Implementation must maintain compliance with HIPAA and other healthcare privacy regulations.
Workflow integration – The system should reduce rather than increase the cognitive load on providers, requiring minimal additional steps.
Early real-world evidence supports the potential of this approach. A 2023 pilot program at Mount Sinai using AI for message triage reported 85% patient satisfaction when responses were human-vetted, suggesting that the human-in-the-loop model can maintain trust while improving efficiency.
Q: What additional benefits might AI-assisted patient communications offer?
Beyond the primary advantages of time savings and improved response quality, AI-assisted patient communications could deliver several additional benefits:
Enhanced documentation – AI systems could automatically document patient-provider interactions in the EHR, reducing the need for manual entry and potentially improving care continuity.
Improved consistency – Standardized responses to common questions ensure patients receive consistent, evidence-based information regardless of which provider reviews their message.
Multilingual capabilities – AI translation capabilities could help providers communicate more effectively with non-English-speaking patients, addressing an important healthcare disparity.
Proactive patient education – Systems could identify opportunities to include relevant educational resources in responses, potentially improving health literacy and treatment adherence.
Population health insights – Aggregated (and anonymized) data from patient communications could reveal trends in patient concerns, seasonal health issues, or medication side effects, providing valuable insights for healthcare organizations.
Reduced provider burnout – By alleviating the administrative burden associated with patient communications, these systems could contribute to improved provider wellbeing and career longevity, addressing a critical healthcare workforce issue.
Q: What areas need further development before widespread adoption?
Several critical challenges must be addressed before AI-assisted patient communications can reach their full potential:
Managing liability concerns – AI-generated responses may inadvertently create impressions of culpability or liability that aren’t factually present. For example, a system might generate technically accurate but overly cautious language about medication side effects that could be misinterpreted as suggesting the provider made an error. Healthcare organizations need sophisticated review protocols and AI training that recognizes these subtle nuances in medical communication.
Context awareness limitations – Current AI systems may miss important contextual factors in a patient’s history or fail to connect seemingly unrelated symptoms that would be obvious to an experienced clinician. This gap could potentially create medical risks if providers rely too heavily on AI-generated drafts without thorough review.
Legal and regulatory frameworks – The regulatory landscape around AI in healthcare is still evolving. Healthcare organizations need clarity on liability, data ownership, and compliance requirements before making significant investments in these technologies.
Clinician resistance – Despite potential benefits, some physicians may resist what they perceive as further technological encroachment on the patient-provider relationship. Implementation strategies must address these legitimate concerns to gain broad adoption.
Overcoming these challenges will require collaboration between technology developers, healthcare providers, legal experts, and patient advocates to create systems that enhance rather than compromise the quality of patient care.
Q: What’s the likely future of AI in patient communications?
The question isn’t whether AI will play a role in patient communications, but rather how quickly and effectively it will be implemented. The technology is advancing rapidly, with healthcare-specific models showing improved understanding of medical terminology and concepts.
The most successful implementations will likely:
- Start with clearly defined use cases (e.g., appointment scheduling, medication refills)
- Maintain transparency with patients about the process
- Continuously improve based on provider feedback
- Gradually expand to more complex communications as the technology matures
Healthcare organizations that thoughtfully integrate AI into their communication workflows stand to gain significant efficiency while potentially improving patient satisfaction through more timely and comprehensive responses.
Bibliography
American Medical Association. (2022). Time spent on administrative tasks in healthcare settings.
Journal of Medical Internet Research. (2023). Physician attitudes toward artificial intelligence in clinical workflows.
Pew Research Center. (2024). Public attitudes toward artificial intelligence in healthcare.
Mount Sinai Health System. (2023). Patient satisfaction with AI-assisted triage in portal communications.
Disclaimer: Our content is intended solely for educational and informational purposes and should not be viewed as a substitute for professional advice. While we strive
for accuracy, we cannot guarantee that errors or omissions are absent. Our content may utilize artificial intelligence tools, resulting in inaccurate or incomplete information. Users are encouraged to verify all information independently.
—
© Copyright 2025 The ADD Resource Center. All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means without obtaining prior written permission from the publisher and/or the author.
ADD Resource Center: Your Partner in Understanding and Growth
Your journey toward enhanced understanding and support begins here.
The ADD Resource Center has established a vibrant community of learners, professionals, and advocates committed to fostering positive change through evidence-based approaches and compassionate support.
Our Comprehensive Services
We offer personalized guidance for individuals, families, and organizations through integrated support systems tailored to their unique needs. Our expert team provides targeted behavioral intervention strategies and delivers specialized assistance to healthcare providers, educators, and industry professionals.
Evidence-Based Resources Within Reach
Access our carefully curated collection of informative articles, participate in transformative workshops and seminars, and engage with our advocacy initiatives to promote understanding and reduce stigma.
Our resources are continually updated to reflect the latest research and best practices.
Take the First Step Today
Join our expanding community and discover the impactful difference that expert guidance and support can make in your journey.
Join our Mailing List to stay updated on our latest resources and events.
Disclaimer: Our content is intended solely for educational and informational purposes and should not be considered a substitute for professional advice. While we strive for accuracy, we cannot guarantee that errors or omissions are absent. Our content may use artificial intelligence tools, producing inaccurate or incomplete information. Users are encouraged to verify all information independently.