At RadNet, AI-fueled prior authorization tech is 99% accurate

RadNet is a national radiology provider with 344 imaging centers employing 8,000 people and 500 radiologists. It has regional networks in New Jersey, New York, Delaware, Maryland, California and Florida.


Due to its size and patient-centric services, RadNet has a tremendous volume of procedures – more than 7 million per year, or roughly 27,000 procedures per day. If only 20 percent of those procedures required authorizations (a conservative estimate), that would leave RadNet staff responsible for processing more than 5,400 authorization requests per day. That is a tremendous drain on employees.

“Prior authorization is incredibly challenging for us,” said Charlie Shaw, senior vice president at RadNet. “An ever-increasing number of procedures require authorization, and that authorization is difficult in many instances due to lack of clinical documentation from the referring physician. We often do not get the information needed to process the requests.”

On top of the payer’s proprietary guidelines for authorization, RadNet is expected to have very specific information for each request:

  • Patient’s full name as it appears on the insurance card.
  • Active primary, secondary and/or tertiary insurance information.
  • Referral information and justification for procedure.
  • Referring provider’s clinical notes.
  • Complete patient history.

However, what RadNet often gets is:

  • Incomplete patient demographics, referral information and procedures.
  • Potentially outdated or inactive insurance information.
  • Delayed receipt or complete lack of clinical notes and/or patient history.
  • Evolving and demanding payer guidelines for authorization.

All of these factors lead to delays in authorization in addition to manual effort of following up on static web portals, which leads to patient rescheduling and attrition.


Because time-consuming phone calls to insurance companies were the norm, RadNet sought a solution to help with prior authorization processing.

“We started working with Infinx and they implemented their Prior Authorization Software – software that uses automation and artificial intelligence backed by prior authorization specialists that help with complicated cases and payers that do not accept electronic submissions,” Shaw explained. “Infinx provided a structured, end-to-end preauthorization solution across the entire network.”

Through the platform, RadNet was able to automate job creation, work allocation, resource management and billing functions. It also could effectively track the process with real-time alerts and e-mail notifications. Additionally, it has full auditing control and visibility to help further streamline its prior authorization process.


There are some vendors on the market today that offer automation of prior authorizations. These vendors include CenterX, Klara, PracticeSuite, PriorAuthNow, The SSI Group, Surescripts and ZappRx.


Through integration of Infinx’s authorization determination software into RadNet’s existing systems, RadNet plans to gain the ability to identify authorization requirements at the time of scheduling. RadNet also receives support for requesting clinical information in a timely manner, which assists greatly with case initiations for all payers, Shaw said.

“Infinx now uses artificial intelligence to determine if prior authorization is required,” he explained. “As insurers keep changing their policies, the machine learning algorithms learn and adapt to these changes. The determination will allow RadNet to schedule the patient as soon as possible, which will improve the overall patient experience.”

RadNet also is working with Infinx to develop machine intelligence capabilities to be notified if the prior authorization request requires a clinical review and help predict the turnaround time for a specific case so the provider can give the best possible schedule to its patients.

“If a given procedure requires prior authorization, we can electronically submit the request to payers using the integration with Infinx, reducing the need to make multiple calls or send documents to different insurance companies,” Shaw said. “This eliminates the need to log into multiple individual payer portals.”

The authorization requests can be electronically transmitted using the integration, and Infinx prior authorization specialists are available to perform manual submissions in complex scenarios or if a payer does not support electronic submissions.

Once the authorization requests are submitted, RadNet is provided with an authorization reference number and Infinx keeps following up with payers using an automated process that provides real-time updates as soon as a payer gives approval.

“The solution in the future will continuously evaluate the follow-up against the scheduled date of service and turnaround risk, and we are able to engage Infinx’s prior authorization specialists if follow-up with the payer is necessary to ensure authorization before service is rendered,” Shaw said.

The platform supports RadNet by eliminating the administrative burden of obtaining prior authorizations, he added. Additionally, the platform provides insights that helps RadNet improve the patient procedure scheduling experience without the financial risks of performing procedures without insurance company approval, he said.


As mentioned earlier, RadNet believes roughly 20 percent of its 7 million procedures per year require authorization, which means 5,400 authorizations could be required per day.

Since it started using the new prior authorization platform:

  • The volume of procedures handled by Infinx for RadNet has been 13,000 authorizations per month.
  • The percent of cases that are determined to not require authorization upon review with the payer is 55 percent (done both using artificial intelligence and Infinx specialists).
  • Based on pilot results with some of the RadNet locations, AI-based prior authorization determination engine accuracy is 99 percent.


“The most important aspect of the relationship has been the ability to communicate with the vendor – having a partner that understands the industry problem of prior authorization and its impact on radiology and imaging providers and that delivers solutions with next-generation capabilities including AI and automation to enable our business goals,” Shaw said.

RadNet believes that to manage prior authorization, one needs to combine automation, artificial intelligence and human expertise to completely cover all required prior authorizations.

“Any provider network looking to completely automate solutions for prior authorization should instead look to a solution that does not remove the human element factor,” Shaw advised. “AI can be wonderful, but it has not yet evolved to completely replace the human element to the prior authorization process.”

Twitter: @SiwickiHealthIT
Email the writer: [email protected]

Source: Read Full Article

Create Account

Log In Your Account