In a recent Talking Benefits episode from the International Foundation, host Justin M. Held spoke with Michael Stoyanovich, vice president and senior consultant in Segal’s administration and technology consulting practice, about the use and practical application of artificial intelligence (AI) in health plans.

Defining AI

AI is essentially the use of machines to stimulate human intelligence to perform tasks like “learning,” “reasoning,” “decision making,” and “summarizing.” Its key components are large language models (LLMs) and generative pretrained transformers (GPTs).

An LLM is a type of AI that is trained on massive amounts of text to “understand” and generate human language. It uses advanced statistical techniques and programming to recognize patterns in language to produce fluent, coherent, context-aware responses to questions or prompts. GPTs are the systems we use to interact with LLMs.

Uses Within Health Care Plans

While Stoyanovich notes that AI adoption is not equal in every market, or every organization for that matter, he does see it being used for a variety of helpful functions within health plans. Some common uses include the following.

• Claims adjudication: Using AI to automate claims approvals and denials can reduce human intervention while increasing processing speed and efficiency. AI can also be used to predict and flag claims that may need manual intervention.
• Predictive analytics: AI can analyze health indicators and claims history to identify employees most likely to need medical attention in the future. This enables proactive health interventions, reducing costs in the long run.
• Automated fraud detection: Organizations can utilize AI to detect unusual patterns that may signal potential fraud. This not only reduces financial losses but also helps organizations maintain compliance and regulation.
• Enhanced security: AI can use pattern recognition to identify suspicious network activity and implement advanced encryption for sensitive health and financial information.
• Participant-facing uses: AI can be used to assist with the benefits enrollment process and ongoing participant communications. Some organizations have implemented virtual benefit advisors to answer questions and suggest options and personalized plan recommendations.

Challenges and Ethical Considerations

Like any emerging technology, there are myriad challenges we must face if we want to reap the benefits. It is important to remember that AI can be wrong and produce hallucinations, so it absolutely needs consistent human oversight, management, governance, and review. Stoyanovich emphasized the importance of having good, high-quality data to feed these systems, saying that “if it’s fed with garbage, it’s going to produce garbage.” Another major challenge for organizations hoping to utilize AI is integration. In order to get impactful outcomes from this technology, it needs to be integrated into multiple systems, processes, and workflows—It can’t work on an island. You also have to teach employees how to properly use this technology, especially how to digest and evaluate its outputs. AI requires continual training and upskilling for staff, as well as leadership buy-in.

In addition to organizational challenges, there are also legal and ethical considerations that come with AI use. This is a dynamic environment, so there is still all the same privacy and security guidance at multiple government levels that applies to this data and technology, but also new regulations that continue to emerge. In the U.S., the Centers for Medicare & Medicaid Services (CMS) has proposed guidelines for AI use in health care, and the Health Data, Technology, and Interoperability (HTI-1) final rule mandates transparency for AI algorithms. In terms of ethics, consider the inherent bias within AI outputs and ensure your usage remains transparent and explainable. Plan participants want to know how you are using AI, what you are doing to govern it, how you plan to limit hallucinations, and how you plan to limit the impact on individuals and the organization.

Want to learn more and hear select case studies of AI implementation in the health care industry? Listen to the full episode of Talking Benefits wherever you find podcasts.