Artificial Intelligence (AI) offers tremendous potential to improve patient outcomes, reduce disparities, and make health care more efficient. Realizing this potential requires thoughtful implementation, ongoing evaluation, and commitment to strong ethical standards.
Kaiser Permanente is leading the way with a responsible patient-centered approach focused on trust and clinical excellence. Every AI tool we deploy undergoes a rigorous vetting process to ensure it is safe, effective, and equitable. Our doctors and care teams remain at the center of clinical decision-making, supported by AI tools that enhance — not replace — their expertise.
Kaiser Permanente’s responsible AI approach ensures that the AI technologies we use are aligned with our mission to deliver high-quality, affordable care to our members and communities. Here are 3 examples of how Kaiser Permanente is using AI innovation:
AI ambient listening tool
Clinical documentation has become a significant pain point in modern health care, with many providers working after hours to finish notes. This administrative burden contributes to physician burnout, impacting physician well-being, quality, and safety of patient care, and workforce stability.1,2 A national survey found that among physicians reporting symptoms of burnout, approximately 75% identified clinical documentation as a contributing factor.3
Abridge is an AI ambient listening tool that, with patient consent, transcribes patients’ health concerns during in-person visits. This allows physicians to focus more on the patient during the visit and less on documentation. After the visit, the tool creates a draft clinical note that physicians review and add to the patient’s health record.
Physicians have reported positive experiences with the AI scribe, with 84% noting better visit interactions and 82% reporting improved work satisfaction.4 By reducing cognitive load from typing and providing real-time support during encounters, the technology offers meaningful improvement in day-to-day clinical practice.5
Social Risk Score
Access to adequate nutrition, financial stability, and housing impacts our overall health and well-being. 68% of Kaiser Permanente members, including 90% of Medicaid enrollees, face at least one unmet social need.6
These social needs are closely linked to higher rates of hospitalization, emergency department visits, and mental health care.7 To address this, a Social Risk Score algorithm was developed to identify members most likely to be at risk for food insecurity, financial stability, or housing challenges and connect them to support services. In 2024, nearly 1.3 million members were contacted by phone, text, or email and offered support.
A pharmacy pilot program utilizing an earlier version of the social risk score algorithm found that flagged individuals were twice as likely to report unmet social needs. The most common needs included transportation, financial challenges, caregiving support, mental health concerns, and food access.8 Members who accepted referral support services were more likely to take their medication as prescribed one year after being screened.9
Advance Alert Monitor (AAM)
Hospitalized patients in any health system may suddenly deteriorate, leading to an intensive care unit (ICU) transfer or emergency resuscitation, which increases the risk of severe illness and death. To prevent this, care teams must promptly recognize, communicate, and respond to changes in a patient’s physical and mental health.
Using a predictive algorithm, we identify patients who may be at risk of clinical deterioration while in the hospital. By analyzing electronic health data and alerting appropriate care teams, the system helps prevent problems before they escalate. The AAM provides clinicians with up to 12 hours of advance intervention period, enabling more proactive and personalized interventions. When a patient’s risk exceeds a threshold, a virtual team of specialized nurses reviews the case and collaborates with bedside staff to adjust the care plan.
Operating in 21 Kaiser Permanente hospitals across Northern California, the program saves approximately 500 lives each year.9 As a result of this monitoring tool, patients have lower intensive care unit admission rates, shorter hospital stays, and lower mortality rates.10
We also encourage policy leaders to promote responsible AI through a focus on effective AI design and monitoring, well-designed governance frameworks and guidelines, and coordinated, industry-specific oversight With a commitment to responsible AI, we can help shape a health care system that is increasingly effective, better coordinated, and more affordable. To explore policy opportunities related to AI, visit our page with additional recommendations.
References
Full citations for this document can be found at: kpihp.org/references-ics.