Our work covers a broad range of aims centered on the development and integration of artificial intelligence (AI) technologies that solve important, practical problems for patients, providers and health systems.
We work with clinical, operational, and technical teams to advance the development of clinically relevant models, leveraging quality improvement, implementation science, design thinking, and traditional research methods.
Recent News & Publications
Predicting Avoidable Healthcare Utilization: Practical Considerations for AI/ML Models in Population Health
Mayo Clinic Proceedings, April 2022
In the United States, one in every ten dollars of total hospital expenditures is spent on potentially preventable conditions. Preventable hospitalizations and ED visits add up to $100 billion in cost to the US health care system each year, which is more money than the GDP for 140 countries. Up to 13% of adult hospitalizations and 8% of pediatric hospitalizations are potentially preventable, and the majority of these preventable hospitalizations are due to poorly controlled chronic conditions. As the paradigm shift from fee-for-service to value-based care continues to accelerate, there is growing excitement in using AI/ML to predict clinical deterioration in the primary care setting and reduce preventable ED visits and hospitalizations. In this article, the authors describe five important lessons they learned about translating and implementing AI/ML predictive models to achieve the holy grail of population health: combining accurate predictions with effective interventions to engaged patients.
Using AI to Empower Collaborative Team Workflows: Two Implementations for Advance Care Planning and Care Escalation
NEJM Catalyst Innovations in Care Delivery, March 2022
To facilitate the development of ML models in care delivery, which remain poorly understood and executed, Stanford Medicine targeted an effort to address this implementation gap at the health system by addressing three key challenges: developing a framework for designing integration of AI into complex health care work systems; identifying and building the teams of people, technologies, and processes to successfully develop and implement AI-enabled systems; and executing in a manner that is sustainable and scalable for the health care enterprise. In this article, the authors describe two pilots of real-world implementations that integrate AI into care delivery: one to improve advance care planning and the other to decrease unplanned escalations of care. While these two implementations used different ML models for different use cases, they shared a set of principles for integrating AI into care delivery. The authors describe how these shared principles were applied to the health system, the barriers and facilitators encountered, and how these experiences guided processes for collaboratively designing and implementing user-centered AI-enabled solutions.