2025: AI Transforms Healthcare Operating Models

By Venkatgiri Vandali, president of healthcare and life sciences, Firstsource.

AI is a natural fit in multiple functions across the health plan value chain. In 2025, more plans will deploy generative AI in claims operations, the contact center, quality assurance, training and more.

These areas will become more efficient. In addition, health plans that use AI tools to solve perennial pain points, make employees more productive and deliver better member and provider experiences will start to pull away from competitors.

Why are generative AI and related tools such as machine learning able to deliver more value?

In the past, it took a long time for people to refine their requests to get the data they required. It often meant asking IT for help and could take days. Now AI makes it possible to essentially talk to machines and receive content in real time. This creates tremendous efficiency that will lead to cost reduction and ultimately more value for payers and members.

As an example, generative AI tools quickly extract relevant content from emails, claims, contracts, medical records and more. That minimizes the need for staff to look up information, expedites service processes and increases opportunities for automation.  These efficiencies can net considerable time and cost savings while improving services.

What are AI copilots and AI agents and how do they help?

AI copilots work alongside humans, looking up information, flagging potential errors and suggesting next best actions. AI agents run in the background. Multiple autonomous AI agents can work together to exchange information and act based on business rules in “agentic workflows.”

An AI copilot could oversee an agentic workflow in which AI agents verify member data on a claim is accurate; check and correct edit codes; then evaluate whether the claim will adjudicate successfully. If the agentic workflow determines there’s an issue with the claim, the AI co-pilot will flag the potential issue for an expert human associate to mitigate.

What are some of the major efficiency gains payers can expect from AI?

Generative AI has the ability to manage massive amounts of data and yet also find a needle in that data haystack. This capability did not exist before generative AI and is invaluable to any operation that requires analytics and where resolution speed is critical. For example, things can go wrong in five million claims for a variety of reasons. Before AI, a payer would have to retroactively identify and correct the issue. Today, generative AI can constantly be talking to the system, monitoring claims, following up, ensuring claims are routed and processed correctly. Generative AI can also recognize when something is wrong. Then it routes the claim back to human stakeholders and alerts them to the problem.

AI co-pilots can improve auto-adjudication rates and accuracy. For each claim, the co-pilot can identify the provider contract, the applicable health policy and benefits. Then it can calculate the allowed amount, compare that to the billed amount and adjudicate the claim — all without manual intervention.

Similarly, claims adjudication AI agents and copilots can predict and resolve potential pended claims during adjudication; make intelligent real-time claim edits; use member life events and claims data to predict potential COB in real-time; flag potential appeals and grievances claims; and check for over and under payments.

How can generative AI improve prior authorization and appeals?

AI co-pilots can coordinate agentic workflows that review and validate prior authorization requests and triage incoming appeals. The co-pilot can validate the request against member and provider records, benefit plans, payer guidelines and standard operating procedures. It can summarize medical notes and present the information to a human associate. When additional supporting information is required, a co-pilot can generate an appropriate letter to the provider.

What benefits does AI bring to training and continuous learning?

The biggest differentiation generative AI is bringing to training and continuous learning is personalization. Current training models, even self-directed learning tools, take a cookie-cutter approach that assumes everyone learns the same way at the same speed. But people learn at different speeds and in different ways. Generative AI enables people to learn the way they learn best, whether that’s visual, text-based, voice, video or some combination. That increases speed to competency, whether for contact center work or implementing new care management guidelines.

Generative AI can help launch training programs quickly. AI co-pilots can transform written content into a multimedia, gamified training course with videos in as little as one week. Generative AI tools can read and summarize the key points of a training manual or standard operating procedure and draft presentation decks and video storyboards in a day. Then human experts can review, verify and finetune the content of the AI drafts. A training course can be ready in a fraction of the time it’s traditionally taken to produce videos and custom material.

In the contact center, generative AI can audit 100% of the member-agent interactions, not just one or two percent. The technology can help individual agents understand precisely how to improve the quality of their conversations with members.