Startup Anomaly rolls out AI tech to help predict potential insurance claims denials

Denied insurance claims are a headache for providers and payers and ultimately impact patients as well.

Denials and associated rework cause payment delays and contribute to burnout amongst overstretched providers and staff. Payment delays on denied services, when a payment is ultimately made after the denial is resolved, have a profound effect on cash flow, requiring an average of 16.4 more days to pay than claims that have not been denied, according to a study by Crowe Horwath LLP.  The expense and cash flow implications of this average delay are equivalent to at least 1 percent of a provider’s cost structure. 

It's estimated that 50% to 65% of denials are never worked, resulting in lost revenue for providers.

At the same time, the issue of denials and appeals drives unnecessary administrative costs for payers and abrasion with their provider network and members.

Anomaly, a startup using artificial intelligence to streamline the healthcare billing and payments process, developed a tech solution to tackle this problem. The company's AI technology, called smart response, gives providers a real-time response predicting actionable denials and payment amounts—so providers can correct and avoid denials before they occur.

The software operates directly within providers’ native workflows by integrating with payers, practice management software and claims clearinghouses, Jacob Shiff, co-founder and CEO of Anomaly, said in an interview.

"Anyone who's interacted with the healthcare system knows that billing and payments are really just a complete mess, a tangle of denials, appeals, overpaid claims that are clawed back and underpaid claims that are upwardly adjusted. That system can be really painful for everyone, doctors, insurance companies, and then, above all, patients and it impacts the affordability of healthcare itself," he said.

"Anomaly is focused on eliminating that friction and making healthcare payments work better for everyone. So we've developed some powerful AI-enabled capabilities that analyze billions of transactions in the claims flow to predict and prevent payment-related issues before they occur," Shiff said.

Industry reports and Anomaly’s analysis have found that more than 10% of claims are typically denied, a growing problem tied to increasing billing complexity and disruptions such as COVID-19. While this causes problems for both providers and payers, ultimately patients are often caught in the middle, causing added stress from denial notices or even balance bills.

Founded in 2020, Anomaly’s platform uses machine learning to search for irregularities in medical billing to prevent overpayments and billing errors.

According to a 2019 report published in JAMA, such payment errors make up more than $300 billion annually in unnecessary spending, about 10 cents to every dollar spent.

“Imagine if that was the case with your credit card—if every month you got your statement and 10% of your charges were just incorrect,” Shiff told Fierce Healthcare last fall. “That’s really the status quo in healthcare today.”

The average cost to rework a claim has been pegged at more than $25 based on industry reports from MGMA, PNC Financial Services, CMS, The Advisory Board Company and HFMA. Denials cost payers an average of $6 to $20 per denial in administrative costs, including appeals and support calls, according to Anomaly's research.

Anomaly has focused on using its AI claim prediction engine to develop overpayment prediction, which enables payers to predict and prevent overpayments by learning from previously overpaid claims, and instant payments, which will enable providers to immediately get paid upon claim submission. 

In October, the company raised $17 million in series A and seed rounds to expand the team, develop its technologies and partner with more payers, the company said.

The company is headquartered in New York City, and is backed by Redesign Health and investors RRE, Link Ventures, Madrona and Declaration Partners.

Anomaly's smart response technology is focused on providers to help predict where a potential denial will come from, giving them the ability to correct it before submitting. Anomaly has also seen significant interest from payers interested in deploying the software across their entire provider network in order to reduce avoidable denials and rework before the claim even reaches them.

The company's AI claim prediction engine analyzes thousands of parameters and billions of claims to learn payer-specific rules and nuances for each provider. By continuously analyzing data in real-time, the smart response technology stays up-to-date and automatically adapts to changing trends and new payer rules. Smart Response accurately predicts claim line denials and reasons with over 97% precision, based on an analysis of over $100 billion of billed charges through June 2022, according to Shiff.

The engine is focused on actionable denial scenarios and is able to achieve this level of precision for up to half of total denials (recall) across thousands of payers, all 50 states, and all lines of business (commercial, Medicare and Medicaid), according to the company.

The company currently has beta partners using the product including claims clearinghouses, multi-specialty provider organizations and regional payers, Shiff said.

Anomaly plans to extend its engine to predict final payment amounts and patient responsibilities, which will further increase transparency in the payment process for all stakeholders.

"What we plan to do is build out this intelligent healthcare payments layer to enable fast instant, accurate healthcare payments and that, ultimately, reduces healthcare costs and complexity for everyone," Shiff said.

There are many companies focused on using technology to address billing and healthcare payments. Shiff says Anomaly's AI claim prediction engine offers a differentiated approach to the problem.

"Many players in the industry are using a more traditional rules-based approach. Our AI-driven approach, when applied very intentionally, has benefits over those approaches, especially when we think about dynamically learning the payer and provider-specific nuances," he said.