Using AI to Improve Clinical Outcomes - Cutting Through the Hype

Two recent publications speak to implementing AI in healthcare the right way. A group from several US academic medical centers published on The clinical artificial intelligence department: a prerequisite for success in BMJ Health and Care Informatics. Their contention is that physicians have come to hate computers due to the challenging way EMRs were implemented, particularly that these disrupted their workflows and added burden to documentation.

The EMR deployment was based on corporate and administrative benefits without requiring any demonstrable improvements in processes or outcomes for our patients or ourselves.

They strongly recommend implementing AI the right way, specifically, creating a centralized department in a health system with physicians involved from the start.

The second publication is a new book in HIMSS Book Series by Tom Lawry. AI in Health. He notes not only use cases which demonstrate creating value with AI but stresses the importance of engaging clinicians in the roll out of AI as a strategic asset. He emphasizes how AI will "improve processes such as establishing a prognosis." AI will also "displace or enhance certain types of repetitive work' especially in radiology and anatomic pathology. His recommendations for clinical engagement include: setting goals to promote a cultural shift, talk about rewards and benefits, include clinicians in decision making, provide education about AI and define KPIs for early pilots.

So adoption of clinical AI needs to acknowledge the failures of other clinical IT projects and approaching this one in a more engaged approach. The hype around AI needs to be tempered and molded into a clinical approach which demonstrates value to clinicians in order to be widely adopted.