Google Releases AI Tools To Speed Up Health Insurance Preapprovals

Google on Thursday announced a suite of artificial intelligence tools to help speed up one of healthcare’s big administrative bottlenecks: prior authorization.

This is the process where a doctor needs to file a request with a patient’s health insurer before moving ahead with certain tests, imaging, surgeries or prescriptions. It has remained a largely manual process on both sides with humans sorting through the unique details of a patient’s health plan and medical history across faxes, PDF documents, phone calls and online portals.

The idea of getting health information to flow between patients, providers and insurers may sound basic, but exchanging data across different systems, known as interoperability, remains one of the biggest challenges in healthcare. A consequence can be delays in people getting the care they need, which can exacerbate health conditions and lead to worse outcomes. This is one reason why the federal government has proposed new rules to speed up prior authorization in certain government-funded health plans that would require urgent requests to be processed within 72 hours and regular requests in 7 calendar days – twice as fast as the existing requirement – by 2026.

The software from Google’s cloud unit, called Claims Data Activator, is focused specifically on helping health insurers clean up the data in whatever form it comes in from healthcare providers with an eye towards speeding up the prior authorization process. It relies on a combination of several existing Google AI models that can extract information from documents and organize it in a standardized format to make it easier and faster for the human reviewer. The tools will also allow health insurers to analyze the data and to share information electronically with doctors and hospitals.

Google says these tools are geared towards simplifying the process, not giving recommendations. “We are not automating a yes or no,” says Amy Waldron, director of global health plans strategy and solutions at Google Cloud. “Our whole goal is to shrink the amount of time of getting the information from the provider to the reviewer and making the best use of that reviewer’s time.”

As part of the broader “Claims Acceleration Suite,” Google is recommending tools from two other companies – Myndshft and Pega – that have built solutions using Google Cloud. Myndshft’s software helps doctors and hospitals check a patient’s insurance benefits and submit the prior authorization claim to the insurer. Pega offers workflow software that helps insurers manage the prior authorization review process. “We don’t want to sit there and compete with existing systems that the healthcare industry is using,” says Waldron.

The software tool kit is part of a broader push among big technology companies to jockey for healthcare cloud customers. Waldron says Google Cloud serves “six of the ten largest health plans,” but the company does not disclose its total healthcare cloud customers. In the broader $227 billion cloud market, Amazon Web Services holds around a 33% share, according to Synergy Research Group, while Microsoft Azure comes in around 23% and Google Cloud around 11%.

Some of Google’s previous healthcare AI-related efforts have generated controversy. In 2019, The Wall Street Journal first reported on “Project Nightingale,” a Google partnership with the hospital system Ascension, including using AI to analyze millions of medical records, which raised concerns about data privacy and security (both Google and Ascension said the work was compliant with federal patient privacy laws). In 2021, the tech giant disbanded its standalone Google Health division but said health-related efforts would continue across the company. At Google’s annual health event in March, the company reported updates including changes to search to help people find information about Medicare and Medicaid, developments in medical-specific large language models and tools to help developers build health apps.

If a health insurer wants to use the new software suite, Waldron says they would hire a consultant (Google Cloud has a partnership with Accenture) to train Google’s AI models on the data that is stored within the customer’s cloud environment. “We don’t access our customers’ data,” says Waldron.

Google Cloud has been piloting the data exchange software with insurers over the past few months, including Blue Shield of California and a subsidiary of U.K.-based health insurer Bupa. “You have to meet the providers and hospital systems where they are because this is a big gnarly problem,” says Lisa Davis, senior vice president and chief information officer at Blue Shield of California, which has around 4.8 million members. The health insurer processes around 1 million prior authorization requests annually, she says, and is hoping to cut the time it takes for human reviewers to manually extract and enter data, so they can focus on reviewing requests. “No one has the same technology infrastructure or data infrastructure across the ecosystem.”

Prior authorization is just one area where Blue Shield of California is trying to cut down on paperwork. Davis says the insurer has been working with Google Cloud for the past two years on developing a system to have near real-time claims processing so patients can know how much they will have to pay at the doctor’s office, rather than waiting weeks for a bill to show up in the mail. As more administrative processes are automated, Davis says humans will remain in the loop. “AI will never be the be-all end-all,” she says. “It is an enabler. It’s a tool. You have to have people involved in the system that provide that oversight and quality of care.”