AI is helping the healthcare industry cut wait times for patients and save thousands of hours on data retrieval. Here’s how it could come to a doctor or hospital near you.

Humans still do the heavy lifting, like medical research, but artificial intelligence is starting to help out in healthcare, too: One insurance company is automating administrative customer-service tasks, while New Mexico is using AI-driven tech to help enroll eligible newborns into Medicaid.


AI can help insurers navigate patient records much faster, one automation company said.
In New Mexico, AI helped parents access newborns’ coverage within minutes, an official said.
Healthcare companies can use AI-led automation to speed up the work of customer-service agents.
This story is part of “How Emerging Tech is Changing Everything,” a series exploring the transformative impact of tech innovations across industries.

It’s standard practice for US patients to call their insurer when their doctor has prescribed them a new medication or test or procedure. They’d need to know — would their insurance cover it? Getting that answer can take days, but it’s one of the data-heavy processes that some healthcare institutions are now looking to speed up with artificial intelligence.

At the insurance company Humana, customer-service agents have been using an AI-driven tool called Automation Co-Pilot, by the tech platform Automation Anywhere, to help with those kinds of questions, said Yan Chow, the global healthcare leader at Automation Anywhere, who had also worked for decades as a pediatrician at Kaiser Permanente.

The agents interact with the technology through a dashboard with buttons, which they can push to retrieve insurance and other information of the customer they’re speaking with, according to Chow.

Healthcare firms are using AI to automate tedious administrative tasks

Customer representatives at insurance firms usually check databases for information like how the patient is insured and whether they would need to undergo any additional tests before the insurer can approve covering a procedure, Chow said.

The Automation Co-Pilot tool can help agents figure out, often within minutes, whether a given procedure is covered for a patient, he said.

“The beauty of having automation do that is because there’s not very much human decision-making,” Chow said. “It’s just fetching information.”

“Otherwise, the customer probably will get an answer like, ‘We’ll call you back, we’ll give you an answer within three days,'” he added. “And that’s not something people like to hear.”

In a case study issued in February, Humana also estimated that automating some of its administrative processes helped reduce 684,000 hours a year in time workers spent dealing with documents.

Humana also said it promoted the automation process within the company as a digital worker called “Allie,” which Forbes previously reported. Humana didn’t comment for this story.

Health institutions proceed with caution amid concerns about accuracy and privacy

Still, generative AI in particular carries significant risks, as OpenAI CEO Sam Altman has often described. Its output can be biased if it’s trained on incomplete data. It’s prone to making up information, known as “hallucinations.” Given the safety and privacy considerations in medicine, AI is being implemented cautiously and more for lower-risk functions like pulling up and synthesizing certain types of data quickly, Chow said.

“Many healthcare organizations are trying to figure out where can AI be used along this spectrum — where is it safe, where does it need to be vetted, where does it need to be tested for a while,” he said.

Meanwhile, government-administered health insurance has also met its own AI intervention.

In recent years, the New Mexico Human Services Department automated the process of enrolling eligible newborns for Medicaid coverage, which the agency previously told Insider takes about 10 minutes instead of the month or so it once did.

The agency calls the process “baby bot,” an effort driven by technology from IBM and SS&C Blue Prism, and used at several hospital organizations in the state and nearby areas.

The success of such tools has led the agency to use AI-driven automation for other administrative functions, like indexing returned mail sent to residents, according to Shanita Harrison, the customer-innovation director for the New Mexico Human Services Department.

That development has helped the agency quickly send information to residents about renewing their Medicare coverage this year, after a measure that Congress passed to preserve coverage earlier during the COVID-19 pandemic expired in March.

Harrison said using tech to speed up indexing returned mail had helped the agency handle a process called “unwinding,” in which it reviews whether residents previously enrolled in Medicaid are still eligible. (The pandemic-relief measure that Congress passed in March 2020 had directed states to maintain uninterrupted Medicaid coverage for a period of time.)

“That is really helping us with the public-health unwinding,” Harrison said.

Doctors and patients may get more valuable facetime with AI taking some of the paperwork

The industry seems poised for a broader transformation, as new AI tools emerge to also cater to individual healthcare professionals.

Doctors, for instance, are using AI tools to help them document discussions with patients and simplify some of the paperwork they typically need to do when patients see them, The New York Times reported.

Amazon Web Services, which is developing AI offerings, just rolled out a feature called AWS HealthScribe, meant to help medical employees document their interactions with patients, according to the company.

“I hope that AI will make healthcare more accessible to the consumer and make healthcare itself more sustainable for professionals,” Chow of Automation Anywhere said. “I’m looking forward to machines serving us, as opposed to us serving machines.”

Read the original article on Business Insider

Leave a Reply

Your email address will not be published. Required fields are marked *