By Matthew Hawley

When it comes to AI’s potential to transform claims management, we’re missing the real story.

There is a lot of buzz about the impact artificial intelligence could make on healthcare, from improving care management to speeding drug development to driving efficiency and revenue. New research estimates wider use of AI across healthcare could save $200 billion to $360 billion annually. Today, revenue cycle and finance leaders are leaning into ChatGPT to optimize coding for cleaner claims on the provider side or accelerate prior authorizations on the payer side.

But AI’s promise in strengthening payment integrity for health plans lies heavily in machine learning. Harnessing this potential will depend on a higher degree of collaboration between clinicians and data scientists and a thoughtful approach to operationalizing machine learning in claims management.

A new frontier
Machine learning isn’t as new as generative AI, but our understanding of how to leverage it to strengthen payment integrity is becoming more sophisticated.

There are two types of machine learning models:

• With supervised models, labeled datasets are fed into a machine to train it to identify similar examples from raw data or make predictions based on the data, such as whether a claim is likely fraudulent. These models also “score” the value of the prediction according to how likely it is to be true.
• Unsupervised models analyze unlabeled data to identify aberrant behaviors. These models demonstrate strong potential to identify fraudulent claims, according to recent research. Moreover, the models get better the more data they consume.

When combined with natural language processing (NLP), a type of AI that derives meaning from written text and verbal discussions, these models speed healthcare claim processing through the use of both structured and unstructured data. For instance, NLP can quickly detect information that supports a medical diagnosis and determinations around medical necessity. It can also spot anomalies that could point to inappropriate billing, such as when the codes do not match the patient’s condition or the volume of codes is unusually high.

There is room for both supervised and unsupervised machine learning models in healthcare claims management as they complement one another. However, unsupervised machine learning in particular is a transformational capability for claims management. It positions health plans to broaden their aperture to gain deeper insights from disparate datasets—and flag patterns that might otherwise be missed, including patterns related to fraud, waste and abuse.

Today, many health plans use unsupervised machine learning to detect claims with the highest likelihood of being overpaid based on a review of the supporting documentation in a patient’s chart. Given the high cost and administrative burden associated with medical record retrieval, unsupervised learning is a valuable tool for minimizing unnecessary work by staff. It also reduces the risk of provider abrasion by focusing only on those claims and/or providers that exhibit a suspicious pattern.

Unsupervised models also can detect instances where a provider’s billing behavior varies sharply from their peers. In one example, a provider billed for more than 20 unique genetic tests per member—well over double the average of the provider’s peers. Without machine learning, these types of fraudulent activity would be more difficult to detect. Humans would be left to write queries to spot this activity—assuming they knew the type of behavior to look for—whereas machine learning can uncover this behavior automatically with a high rate of precision.

The path to transformational value
Machine learning is a vital tool for ensuring clean claims are paid quickly and suspicious claims—those exhibiting high potential for fraud, waste and abuse—are detected before payment is made. It increases operational efficiency by accelerating the claim decision-making process. It also reduces the administrative burden for staff—and the likelihood of burnout—by leveraging NLP to review thousands of pages of medical records within seconds, a task that could take a human reviewer hours or days. Moreover, NLP never gets tired of performing the same action over and over. This ultimately makes staff review far more effective and efficient.

But machine learning models in and of themselves do not guarantee payment integrity prowess. Taking a transformational approach to machine learning-driven involves four key elements.

1. Vast amounts of data. Machine learning models must be trained using large quantities of data before they can begin to produce highly effective—and highly accurate—results. The more data that is used—including historical data that reveals trends over time—the greater the statistical likelihood of a correct result. Over time, refinement of the model will depend on continuous access to a data pipeline that feeds the model so it can catch new schemes as they emerge and, just as important, better spot instances where human review isn’t needed.

2. Tight collaboration between clinicians and data scientists. AI and machine learning are valuable tools in a plan’s payment integrity portfolio, but only when backed by significant clinical and analytical expertise. Moreover, the level of collaboration needed between these two groups of specialists will be higher than what health plans traditionally have put into play. That’s because effective use of machine learning requires that health plans ask the right questions of unsupervised models. Much like hospitals that are looking to hire experts to create ChatGPT prompts—for example, “Explain diabetes in simple language for a patient who has just been diagnosed”—machine learning models also need the right prompts to make data analysis more meaningful. Further, once the model detects suspicious activity, it will take a reviewer with special investigative skills and techniques to validate larger patterns of potential fraud, waste and abuse. This is an instance where both clinical and data science skills could prove essential.

Clinicians will also be vital to provider relationship management, assuring providers that even with the introduction of machine learning, payment decisions will still be made based on documented payment policy and clinical guidelines.

3. Data pipelines and tools that empower health plans to dig deeper. Machine learning models that rely only on data that originate within a single health plan will be more limited in their effectiveness than those that incorporate industry benchmark data as well as data across lines of business, geographies and provider networks. Just as these models require a continuous internal data pipeline, an external data pipeline will also be critical in spotting aberrant behavior by specialty, geography, type of organization and other factors, strengthening the health plan’s response. This ensures machine learning efforts are not performed “in a lab,” siloed from other business applications, but rather have a pipeline of rich production data to incorporate and learn from.

4. Project management. Such expertise will be critical to operationalizing work around how to leverage data science and clinical expertise to prevent fraud, waste and abuse. It will also reduce the risk of administrative burden for staff and protect member and provider satisfaction. For example, when machine learning identifies abusive billing patterns, a project manager can help close the loop with a combination of provider education and prepay analytics that spot inappropriate claims moving forward and ensure these claims are not paid. This enables the plan to more rapidly deploy solutions that eliminate breakdowns in payment integrity.

Going beyond the hype
There is a great deal of value to be gained from AI for claim payment integrity if health plans focus on the right areas. Separating the hype around generative AI from where demonstrated success exists—the use of unsupervised machine learning models to protect payment integrity—offers a solid path toward transformational value for health plans and ultimately enhances care for their members.

About the author: Matthew Hawley is executive vice president, payment integrity for Cotiviti.