The health insurance industry increasingly is utilizing artificial intelligence and similar technologies to streamline operations, train employees and enhance customer service.
Insurers have been using so-called traditional AI to process claims, identify fraud and predict risk for years. Now, companies such as UnitedHealth Group, CVS Health, Oscar Health and Florida Blue are experimenting with generative AI models that use context to answer prompts and have great potential in healthcare settings.
Related: Dr. ChatGPT: A guide to generative AI in healthcare
The healthcare industry is ripe for genAI because of its ability to translate formal language, such as disease codes, into natural language, and vice versa. The technology became all the rage when OpenAI released the public version of its AI-enabled chatbot ChatGPT in 2022, and the healthcare sector has shown considerable interest.
"In traditional AI, no matter how complex an algorithm is or how complex an automation is, the machine is doing what we tell it to do," said Dr. Bill Fera, principal at Deloitte Consulting. "With generative AI, machines are actually assigning their own context to information to answer queries and prompts that are designed by a user,” he said.
“There is a lot of potential to improve healthcare processes and the overall healthcare experience and outcomes,” Fera said.
Florida Blue plans to roll out an internal ChatGPT-like tool for its employees next month, said Svetlana Bender, the nonprofit insurer's vice president of AI and behavioral science. “We want to make a difference in how we work. We want to create efficiencies. We want productivity. We want to automate as much as possible so we can spend more time on helping out our members,” she said.
The proliferation of these tools has provoked questions about the future of staffing and the appropriate role of AI and algorithms in medical care. But AI and similar technologies also bring the promise of greater efficiency, administrative simplification and cost savings.
Customer service
Insurance companies are incorporating genAI into customer service. “All of the analytics, scripts and algorithms that are sitting behind call center applications is a very fertile ground for generative AI and AI broadly,” said Mario Schlosser, co-founder and chief technology officer at Oscar Health.
Oscar Health is using the technology to help customer service agents provide better explanations to patients asking about claims, such as why one was denied, Schlosser said.
“We can give the language model the manual that we can otherwise give to a human being to read, and the model can take these two things, put them together and explain why this claim got denied in a very natural language kind of way," he said.
Insurance companies are also using genAI to personalize the member experience.
UnitedHealthcare created a platform to help members find in-network care for specific medical needs via genAI-powered search, according to a spokesperson. “Search results are personalized and ranked, making it quick and easy to find the right care, the first time around,” the spokesperson said.
CVS Health and its Aetna subsidiary has been using AI and machine learning for multiple years and has more recently incorporated genAI. “We have a robust innovation agenda and are leveraging AI solutions to support member engagement and accelerate and simplify our internal processes,” said a CVS Health spokesperson who would not provide details about how the company deploys genAI.
Oscar Health is testing out creative applications of genAI, including its ability to create images. The company is using generative AI to create sketches of in-network providers it plans to show members searching for care. “That stuff would have required tons of graphic design before, and now you can more or less let an algorithm do this,” Schlosser said.
Using genAI tools in member-facing capacities could help insurers retain customers and add new ones while lowering administrative expenses, said Eric Berger, a partner and healthcare consultant at Bain & Company.
“They need to ensure a really good customer experience from the point of onboarding through the course of the care delivery through the payments process,” Berger said. "Improving member experience drops right to the bottom line."
Employee training
Health insurance companies are also using generative AI for employee training programs. Firstsource, a business process services vendor, is working with insurers to develop tools to enhance and expedite training processes.
The company's FirstSenseAI platform, which launched in August, applies genAI to create training curricula based on clients' internal documents “with the click of a button,” said Anup Panthaloor, Firstsource's executive vice president of health plans and healthcare services. Then, it uses third-party tools such as Adobe Article 360 to create videos and simulated training environments.
Firstsource has also developed a platform entitled AI Coach that allows call center employees to engage in simulated service calls with the technology that responds as if it's a customer. Then, the platform assigns a score to the conversation.
“Now, I don't need a trainer sitting with one individual taking a mock call, and I can scale this up. Hundreds of people can do this self-paced learning, and then I can look at the results, and say, ‘OK, this person needs help,’” Panthaloor said.
Incorporating this technology can help companies save on labor costs. What that means for employment over time remains to be seen. “Workforce changes are inevitable, and we've seen that through a lot of automation," Bender said. "What we're trying to do is be proactive about it and upskill our employees."
Safeguards
The opportunities are huge, but the technology has also sparked concerns it may advance misinformation and perpetuate racial biases in clinical practice. UnitedHealth Group, Cigna and Humana each face lawsuits alleging they inappropriately use AI and algorithms to automate prior authorization and claim denials, for instance.
Florida Blue, which began experimenting with genAI a year ago, assembled a center of excellence comprising a small group of employees that monitors how it uses the technology. Given the sensitivity of member health data, the insurer aims to ensure human oversight, Bender said.
Oscar Health is taking a cautious approach to genAI because the technology could produce inaccurate responses if it doesn't have the requisite data.
“The way we solve it right now is we use the models only in situations where we can rely on there being no direct impacts to any real end user if the model hallucinates,” Schlosser said.
Oscar Health and CVS Health are among more than a couple dozen healthcare organizations that signed a White House pledge to safely and responsibly deploy AI in healthcare.
Oscar Health is also monitoring the effectiveness of its generative AI-powered tools. Schlosser said the company found that language models are better at remembering details than customer service representatives, but that the human agents are more adept at incorporating contextual information about specific providers and local geography.
