Artificial intelligence (AI) and machine learning (ML) are making their mark in every industry, and managed care is no exception. A presentation at AMCP Nexus 2023 offered insights into how AI and ML can help managed care plans increase productivity and performance.
The presentation was led by Jessica Hatton, PharmD, BCACP, associate vice president of pharmacy in Ohio for CareSource, and Nick Trego, PharmD, senior vice president of clinical analytics and client services at HealthPlan Data Solutions, Inc.
They defined AI as “the theory and development of computer systems that are able to perform tasks that normally require human intelligence.” This may be divided into “weak AI,” which focuses on specific tasks, such as GPS systems, customer service chatbots, and smart home devices. “Strong AI” is more theoretical and has the ability to equal or surpass human intelligence.
ML is an application of AI, and the presenters defined it as “the use and development of computer systems that are able to learn and adapt without following explicit instructions.” Variations include:
- Supervised learning, whereby labeled datasets train a machine to identify existing patterns. For example, your email account uses this technology to filter out promotions and spam.
- Unsupervised learning, whereby unlabeled datasets can identify unknown patterns. A business example for this is a customer segmentation tool.
- Reinforcement learning, whereby the system is rewarded for identifying the best path or selection and learns from that. This level is used in gaming applications and self-driving cars.
AI and ML are powerful, programmable tools that can increase productivity and performance in managed care, the presenters said. The first area of great potential involves reading and interpreting pharmacy benefit manager contracts. These documents are complex and can exceed 1,000 pages, and current practice is manual review. With AI and ML, models can be trained to identify key elements in contracts and reduce review time to mere minutes, the presenters said.
The second area of potential involves monitoring pharmacy claims, complex documents made even more challenging by their enormous volume. Current monitoring involves manual review and invoice reconciliation, often selectively and restrospectively. AI and ML could be used to identify deviations from a plan as they happen in almost real time. The technology could quickly and efficiently review 100% of claims, rather than a portion. The technology is also being considered to improve member engagement, customer service, and formulary management.
Hatton J, Trego N, Horchani T. Can Artificial Intelligence Help Managed Care Plans Make More Sense of Their Pharmacy Data? Session R1. Presented at AMCP Nexus 2023; Oct. 16‒19, 2023; Orlando, Fla.