AMA Report Outlines Considerations for AI Integration in Healthcare

A report published this week by the American Medical Association (AMA) and Manatt Health outlines the transformational potential and associated risks of augmented intelligence, also known as artificial intelligence (AI), tools in clinical and administrative applications.

The AMA emphasizes that growing interest in AI necessitates increased education for clinicians to successfully navigate the deployment of these technologies.

“The Emerging Landscape of Augmented Intelligence in Health Care” report explores key terms and definitions, potential applications and use cases, and other opportunities and risks that AI tools present.

Emerging AI capabilities identified in the report include identifying characteristics within data; translating data inputs into other formats or data types; summarizing data inputs to make them more accessible; predicting future events based on historical data; and providing recommendations or guidance.

As these capabilities have advanced, some medical specialties have adopted AI more quickly than others. At the time of writing, the United States Food and Drug Administration has approved just under 700 AI- and machine learning-enabled medical devices. Of these, 531 are in radiology, 71 are in cardiology, and 20 are in neurology.

Potential and future use cases also vary significantly across specialties, but the report indicates that real-time clinical transcription; answering patient questions via chatbot; drafting personalized patient education materials; and predicting adverse clinical outcomes are use cases in practice across all medical specialties today.

The report further notes that drafting responses to patient in-basket communications; converting clinical notes and other data into standardized, electronic health record (EHR)-friendly formats; and supporting patient triage are use cases that are either currently in use, but not at scale, or are likely to be in the future across all specialties.

The report also highlights a plethora of non-clinical AI use cases to bolster access to care, administration and revenue cycle, regulatory compliance, patient experience, and other applications.

However, alongside these opportunities, the AMA underscores the many challenges and risks healthcare organizations face when pursuing AI implementation.

Bias, explainability, transparency, model hallucination, coding and payment, privacy, regulation, and liability are concerns that providers must reckon with prior to, during, and following deployment of AI tools. These hurdles require significant investments from stakeholders and government entities across the industry to overcome, and the rapid advance of AI technologies makes doing so particularly difficult.

For healthcare organizations considering implementing AI, the report lays out key questions leadership should be asking during the four stages of adoption: identification of challenges and use cases, evaluation of AI tools, implementation, and management.

Throughout these stages, keeping risks, liabilities, infrastructure needs, financial implications, and health equity top of mind is key.

To further support responsible AI evolution in healthcare, the AMA indicates that it will develop principles for health AI use, support policy development at the state and federal levels, collaborate with technology leaders to help guide AI research, and provide AI resources for healthcare professionals.

The AMA’s report comes as health systems increasingly assess how they can effectively and safely deploy AI tools.

This month, a report from the Center for Connected Medicine (CCM) at the University of Pittsburgh Medical Center (UPMC) revealed insights into how healthcare leaders are navigating the complexities around AI.

The report found that healthcare executives are prioritizing governance and oversight efforts to help them harness the benefits of these technologies while minimizing the potential harms. The findings also highlighted that health systems are increasingly interested in how advancements like generative AI can enhance healthcare.