We pointed a few months ago to the push to get healthcare providers to begin considering the ethical side of AI even as they expand their use of the range of emerging technologies.
The latest step in that direction comes from Utah-based Intermountain Healthcare, which has the announced the establishment of a Data Science and Artificial Intelligence Center of Excellence.
The nonprofit system of 25 hospitals and 225 clinics spread across Utah, Idaho and Nevada says its goal is to ensure ethical AI standards and set goals to further improve patient care.
The organization says its center of excellence will be a new integrated model that brings experts together from multiple disciplines – such as data analytics, applied mathematics and statistics, computer science, behavioral sciences, econometrics, computational linguistics and clinical informatics – along with expertise from various clinical specialties.
“At Intermountain, we utilize machine learning algorithms that emulate human cognition to help providers improve their diagnosis and treatment of medical conditions, predict and identify diseases and infections in their early stages ( including COVID-19 ) and prevent hospital-readmissions,” said Albert Marinez, chief analytics officer at Intermountain.
To date, AI-centered projects at Intermountain have included an ePneumonia protocol that the organization says has saved approximately 1,166 lives per year since 2015, and a program that identifies hospital patients who are at risk of declining, which has led to lower mortality rates.
“We’ve developed an AI Playbook as a framework to deploy and scale human-centered AI that is transparent, equitable, ethical, and above all, ensures patient privacy,” said Greg Nelson, assistant vice president for analytics services at Intermountain. “The playbook outlines goals to establish appropriate AI governance, set validation and documentation standards, detect inherent bias, ensure data integrity and promote AI literacy among caregivers.”
Nelson added that establishing such AI standards is a natural outgrowth of Intermountain’s focus on becoming a model health system and developing ethical healthcare leaders.
Currently, any data product used at Intermountain Healthcare follows a validation process following guidelines based on U.S. and international best practices. Concepts are derived from the U.S. Food and Drug Administration’s definition of validation and require verification and documentation to meet predetermined specifications and quality attributes.
Intermountain caregivers have a long legacy of using computers to improve patient outcomes, which were pioneered in the 1950s at LDS Hospital in Salt Lake City by Homer Warner, MD. Dr. Warner and his colleagues went on to develop one of the nation’s first electronic medical records, and designed a system in the 1970s to assist clinicians in bedside decision-making.
The organization notes that there are more than 130 algorithms that have been FDA approved or cleared as AI-enabled processes for conditions ranging from breast cancer to sickle cell disease to schizophrenia.
Photo by Posteriori/Getty Images