Health care operations professionals discuss the power of generative AI

RSM’s Virtual Health Care Day panel, “Hyper automation: Unleashing the power of generative artificial intelligence in enhancing health care operations,” was an interactive discussion focused on how top health care systems leverage generative AI to address some of the toughest challenges facing hospital operations. A few examples of these challenges include patient portal communications, complex patient chart reviews, and information technology service ticket automation.

The panelists, representing payers, providers, and key deployers of AI solutions, included:

• Corinne Stroum, Director, Consumer Experience & Insights, Scan Health Plan
• Steve Kos, Senior Director, Revenue Cycle Applications, Baptist Health
• Randall Nale, Director, US Healthcare Solutions, Microsoft US Healthcare and Life Sciences
• Tom Bethke, VP of Information Systems Infrastructure, Allina Health

John Bennett, health care consulting director at RSM US LLP, moderated the session.

Evolution of artificial intelligence

AI was first introduced as a concept in the 1950s. Over time it evolved to machine learning, then deep learning, and today generative AI. An example of a generative AI solution is ChatGPT, an AI language model designed to classify text with higher accuracy, enable advanced natural language understanding, and provide customizable solutions—a valuable tool to improve health care outcomes.

“GPT technology isn’t a stand-alone offering; it is like a Lego brick that fits within a larger infrastructure of technology,” said Bennett. “Artificial intelligence is like inserting a human into a production process. In the past, we had a very manual process that involved working with data, labeling data, testing models, and adjusting models to meet business needs, which meant health care organizations had to secure and leverage highly valuable and expensive time from health care professionals.”

AI and compliance in the health care space

While there are many benefits of generative AI, most health care leaders agree appropriate guardrails, due diligence, and protocols are critical to protect private patient information. Platforms must ensure data security and HIPAA compliance. A key element of this compliance is ensuring that a human oversees final decisions. Strong governance is a must and should be assessed frequently.

Health care use cases

The number of AI use cases in the health care industry continues to grow. Today, areas in which companies are using generative AI include:

• Responsible AI governance framework
• ServiceNow ticket triage
• Patient messaging automation
• Patient chart summarization
• Data enrichment, curation, classification and embedding
• Supply chain documents and contract summarization
• Revenue cycle, prior authorization, and claim status
• Call center management and analysis

“One of the things we are trying to do is see how effective message triage can be using artificial intelligence,” said panelist Bethke. “So, how do we get messages routed to the right people? The people that generally can decide that are physicians, so we would ask them to review hundreds or thousands of messages and use that to train the artificial intelligence. One of the biggest challenges was getting their time. Now that we have shifted to current generative artificial intelligence, we can work much faster and more efficiently by asking a lot less of our physicians.”

“We have seen really good success with being able to leverage artificial intelligence for prior authorizations, and we are working on the status of claims as well,” said Kos, another panelist. “There is a tremendous opportunity to depend on artificial intelligence to do tedious work and leave important decisions and actions to the end user.”