Can Artificial Intelligence Deliver in Real-World Health Systems? Early Insights From Augmented Intelligence in Medicine and Healthcare Initiative's 5 Funded Projects.
TL;DR
Early insights from AIM-HI-funded projects to inform real-world AI implementation highlight that implementation success depends on thoughtful integration into clinical environments, strong partnerships with stakeholders, as well as sustained evaluation and monitoring.
Abstract
INTRODUCTION Artificial intelligence (AI) holds tremendous promise to improve clinical decision-making across diagnosis, risk assessment, and patient care. However, most prior work has focused on model development in controlled settings with limited evidence on real-world implementation. The Augmented Intelligence in Medicine and Healthcare Initiative (AIM-HI), led by Kaiser Permanente and funded by the Gordon and Betty Moore Foundation, was established to evaluate and support integration of AI tools into routine clinical practice. This article summarizes early insights from AIM-HI-funded projects to inform real-world AI implementation.
Methods
AIM-HI funded 5 projects through a national, multistage review process using a structured scoring rubric. Projects addressed sepsis management, venous thromboembolism risk assessment, diabetic retinopathy screening, cardiac amyloidosis detection, and pediatric asthma risk prediction across diverse health care settings. The authors synthesized cross-project findings related to implementation processes, challenges, and lessons learned.
Results
Real-world AI deployment was feasible across varied clinical environments. Common challenges included electronic health record integration, data complexity, regulatory requirements, and variation in clinical workflows. Other common themes included stakeholder engagement, local adaptation, quality assurance, and performance monitoring.
Discussion
Findings highlight that implementation success depends on thoughtful integration into clinical environments, strong partnerships with stakeholders, as well as sustained evaluation and monitoring.
Conclusion
Effective AI adoption in health care requires careful integration, stakeholder alignment, and ongoing evaluation. Initiatives such as AIM-HI are essential for building the evidence base for scalable, real-world AI implementation.