https://www.techtarget.com/healthtechanalytics/feature/4-top-use-cases-for-agentic-AI-in-healthcare
AI agents represent perhaps "the biggest untapped opportunity" in healthcare because they work autonomously around the clock, according to Wes Little, executive vice president of analytics and AI at WellSky, a company that offers software, analytics and services to deliver intelligent coordinated care, particularly post-discharge or post-acute care recovery.
Agentic AI consists of AI systems that perform autonomous actions and aid decisions. They also help clinicians when there are staffing shortages. Fertility, for example, is an area that is using AI while it deals with a clinician shortage and increased demand for services, according to Irene Alvarado, cofounder and CEO of Berry Fertility, a fertility management app for IVF, embryo transfer and egg freezing.
Use cases for agentic AI in healthcare include scheduling, authorizations and patient engagement. Agents also access application program interfaces to perform tasks such as querying an EHR, Alvarado explains.
Today's AI agents continue to advance and are now able to read full web pages and understand the information.
"I think we're headed toward an era when the majority of the traffic on the internet will not be humans," Little says. "It'll be agents going out and finding information, collecting it, and then inserting information into other systems."
Here's a look at how agentic AI can help with some key functions in healthcare.
Healthcare organizations use voice agents to help with administrative tasks such as scheduling patient appointments and sending reminders via text. WellSky's voice agent, for example, is integrated into the company's scheduling application and allows providers to call their patients the day before an appointment, which they didn't have the resources to do before AI agents, Little says.
"There's nothing more costly to a provider, both in terms of actual cost and opportunity cost as when that nurse shows up at the house at 9 a.m., and the patient has forgotten they were going to be there," he says. "It's a missed interaction opportunity, and you have to go about rescheduling. That takes a long time."
WellSky's AI agents use generative AI large language models similar to that of ChatGPT and interact with an application from telephony provider Twilio to make a voice agent's functionality possible. In addition, its agents have improved in just the last few months as far as latency and responding to patients effectively, according to Little.
Now AI agents are more advanced because of the ability to have a conversation with the patient, Little explains. AI agents make natural conversations possible compared with traditional automated calls in which a caller simply presses 1 to confirm an appointment or 2 to decline, according to Little.
"Then you have a whole other set of interactions with a [person] who follows up with you," Little says. "Now you can have that whole interaction the first time that patient answers the phone."
Without APIs, AI agents can log in to an EHR to access a calendar to help clinicians fill an open slot, Alvarado says. An AI agent can follow up with patients, send a confirmation email and access SMS reminders, she said.
Other companies that offer voice-activated AI agents to help with appointment scheduling include Amelia and Innovaccer.
Interoperability and APIs are also making agents effective in areas such as authorization, Little says. Voice agents call up payers to seek authorization for a treatment that will be delivered.
"Preauthorizing that care is hugely valuable for providers who otherwise would have to have a big staff calling those payers and following up to ensure the patients get the care that they need," Little says. "Now a lot of that can be done autonomously via agents."
Agents can interact with payer portals and perform "web crawling" to collect the data needed from providers to authorize payment.
AI voice agents can save clinicians time during insurance preauthorization through robotic process automation, Alvarado says.
Because not all insurance covers fertility, for instance, finance coordinators speak with patients to go over the insurance terms, such as medical necessity, Alvarado said. AI agents are just "barely starting" to perform the calls to get a preauthorization from an insurance company.
"In some cases, they're not replacing humans, but they're validating what the humans are researching to save time," Alvarado said.
When patients begin researching fertility treatments, they may not get started for 4 to 8 months, according to Alvarado. "Anything that could reduce that time of finding out your insurance benefits is a huge part of what is useful, both for the patients to get treated faster as well as for the clinics to sort of [fill] that slot faster."
Voice agents can help physicians and patients research which medications are covered.
AI tools aid nurses by writing drafts of notes or providing summaries of patient encounters. Oracle Health's Clinical AI Agent, for example, uses voice recognition technology to record physician-patient interactions and adds draft notes to the Oracle Health EHR, formerly Cerner. Organizations such as Beacon Health System in South Bend, Indiana, uses ambient AI as part of Oracle Health's Clinical AI Agent to fight burnout and help with administrative tasks.
"It improves the trust between the patient and the physician because our notes are more comprehensive related to what the conversation entailed," said Dr. Scott Eshowsky, chief medical information officer at Beacon Health System, in an Oracle case study.
When patients download a patient engagement app, AI agents can check in with patients on how they're feeling and capture valuable information that allows providers to offer guidance on potential risks or issues.
As agents speak to patients every day, they collect data points that can be fed back into a predictive model, according to Little.
"You're increasing the amount of data that you capture, which is inevitably going to make the analytic engines much more performant at identifying the risks that you want to avoid," Little says.
If a patient is having difficulty with a hip and knee replacement after discharge, AI agents can collect this information to avoid the chance of readmissions, according to Little.
"People can slip through the cracks once they've left the hospital doors, and this gives you the ability to stay more connected with those patients than you've ever been able to before," he says.
AI agents can reach out to patients to make sure they understand their upcoming procedures and follow up after, Alvarado says.
If an agent gets in touch with a patient, there's often a "human in the loop" in which humans must review the action before it occurs, according to Alvarado.
As AI agents interact with patients more, they can learn to answer patients' questions more effectively, Alvarado said.
"You can have agents that are self-learning that will take feedback and adjust themselves next time," she said.
Brian T. Horowitz started covering health IT news in 2010 and the tech beat overall in 1996.
22 May 2025