Beyond the Hype: Real AI Use Cases to Drive Payment Accuracy

Table of contents

More than five billion claims are processed in the U.S. each year, making it difficult for health plans to catch all errors and resulting in hundreds of billions of dollars in waste.^1,2 But with recent breakthroughs in artificial intelligence (AI), health plans are exploring more effective ways to leverage emerging technology to drive more efficiencies.

When done right and paired with deep healthcare expertise, AI can support improved value in payment integrity. It can help drive more targeted and precise interventions, scale error detection, augment higher quality and more consistent human reviews and explore new opportunities to identify inaccuracies.

While there is ample marketing and buzz about what different companies offer, it can be difficult to separate flashy marketing from true value. It’s imperative to follow real use cases when determining how AI can drive healthcare payment accuracy.

Use case #1: Scaling error detection

Provider contracts are essential in supporting accurate payment, but they’re frequently changing and often stored in a variety of file formats across disparate systems. This makes it challenging for health plans to catch contract-based errors at scale and can lead to a significant number of inaccurate payments.

At Optum, we’re using large language models (LLMs) to read these large, text-based documents, identify relevant contract terms and convert them into a machine-readable format. We then use this digitized contract data, along with historical claim data, to build AI-powered decision intelligence that directly integrates with data mining systems to expedite the development of contract-based claim edits.

This AI-enabled automation helps health plans scale the detection of contract-based errors and overpayments, helping drive healthcare payment accuracy.

Use case #2: Augmenting human reviews

While AI can be helpful in scaling automation and driving efficiencies, it’s important to keep a human in the loop, especially when making clinical decisions. Once a claim is determined to have a high probability of error, we send it to a human clinical reviewer for further analysis. Oftentimes these reviews are manual and labor-intensive, leaving the process vulnerable to human error and claim processing delays.

To enhance quality and consistency across reviews, we’ve built AI-powered augmentation tools using LLMs to extract specific terminology from medical records that are associated with clinical guidelines, relevant to the CPT code set included in the claim. The tools highlight the potential evidence directly within the medical record so reviewers can more quickly determine whether the medical record coding follows clinical guidelines.

These technologies alleviate the task of sifting through the medial record to find specific terminology and evidence associated with clinical guidelines. They help reviewers avoid missing critical information in the claim and medical record and provide them with real-time insights, enabling quicker, more consistent decisions and improving the efficiency and quality of the claim review process.

Use case #3: Exploring new opportunities to improve accuracy

Another way we use AI in our payment integrity solutions involves leveraging generative AI agents powered by LLMs to identify new ways to detect payment inaccuracies. Through an integrated ecosystem of AI agents, our researchers can tap into a vast repository of analytics to help uncover gaps and opportunities for improvement. For instance, as the healthcare industry transitions from inpatient to outpatient care, our AI agents can analyze relevant data to provide insights and actionable recommendations.

These AI agents empower our researchers to more efficiently explore and validate recommendations by cross-referencing internal knowledge bases, online resources and other sources. By automating routine tasks, our team can dedicate more time to innovative and high-impact projects, accelerating the pace of our research and development efforts toward smarter payment integrity innovations.

While these use cases for using AI to drive healthcare payment accuracy add meaningful value to payment integrity programs, they are just a starting point. We believe AI has the potential to completely transform the way health plans approach payment integrity.

Responsible use of AI

At Optum, we have robust guardrails in place and conduct third-party reviews so you can have confidence that our models are designed to be ethical, transparent, explainable and to do no harm, and will meet the requirements and regulations of healthcare’s robust privacy and security standards.

About Laika Kayani, Vice President, Payment Integrity Product Management

Recently named one of Becker’s 100 Women in Health IT to Know in 2025, Laika leads payment integrity product management at Optum, focusing on technology-enabled solutions that help reduce fraud, waste, abuse and error and improve payment accuracy across the claim lifecycle.