Health Systems Tap GenAI to Transform Patient Care

Two major health care providers — Highmark Health and Hackensack Meridian Health — are betting big on the technology to transform care delivery, reduce burnout and reimagine patient experiences.

During a press briefing Wednesday (April 9) convened by Google Cloud, Richard Clarke, chief data and analytics officer of Pittsburgh-based Highmark, and Sameer Sethi, chief AI officer of Hackensack Meridian, discussed their practical experiences with rolling out artificial intelligence (AI) in a highly regulated industry.

“We really see this as a transformational technology,” Clarke said.

Highmark, which operates both a payer and provider network, has rolled out GenAI tools to more than 14,000 of its 40,000 employees. Already, they’ve surpassed one million prompts, using Google Cloud’s Gemini and Vertex AI as foundational tools.

“It’s been really a cornerstone for us to get comfortable with that, because we’re obviously dealing with incredibly sensitive information and data every single day,” Clarke said.

The impact also goes far beyond efficiency. Clarke said Highmark is using AI for software development, in its contact centers, for prior authorization processes and most notably, in ambient listening tools, which are technologies that use voice recognition to transcribe and interpret conversations between providers and patients.

“That has been a true gift to bring joy back to practice for many of our clinicians,” Clarke said.

Sethi said Hackensack Meridian is taking what he calls a “hyperscale” approach to AI deployment.

The Edison, New Jersey-based health system has grouped its AI investments into six strategic focus areas:

• Personalized and equitable patient and employee experiences

For example, it ensures that employees can get the right information they need at the time they need it for their jobs. For patients, they will have access to and understand their medical records before meeting with the physician, among other use cases.

• Administrative and clinical efficiencies

Automating tasks like note-taking and manual data entry to speed up administrative tasks. For example, rather than nurses manually entering test results from paper documents, AI tools now scan and integrate them into electronic medical records.

• Burnout alleviation

Reducing the administrative burden on doctors and nurses will mitigate burnout.

• Disease prediction

Looking through the patient’s data not only for early disease detection but to predict potential illnesses from signals in their medical records.

• Precision treatment

Using AI to customize treatment to the individual patient level that is attuned to their needs.

• Research and innovation

Finding more ways to advance medical innovations, with help from the latest technologies.

Read more: Microsoft’s AI Healthcare Push Aims to Boost Efficiency, Enhance Patient Care

Beyond current implementations, both leaders are eyeing the next frontier: AI agents capable of reasoning, planning and taking action across systems.

“It’s the ability to go from insight to action,” Sethi said.

Sethi described a future vision where a single agent could handle an entire patient journey. From setting a patient’s appointment, to arranging for a ride to and from the hospital, making sure a wheelchair is available as requested, and picking up the medication en route to bring the patient home.

Those are six or seven steps done by human workers today, Sethi said.

While still aspirational, some pieces are already in place. Sethi said Hackensack has built a “nurse agent” that allows nurses to query vast documentation and receive recommended next steps, replacing time-consuming keyword searches.

Clarke echoed the importance of the transition to AI.

“Most of our work has been internal to our workforce,” he said. But “I’m very excited about this next phase (of AI agents) directly interacting with our patients … where high quality guidance is there all the time, always on.”

Clarke also said that the challenges they thought would impede AI adoption a few years ago, such as hallucinations and costs, “are really just not turning out to be the barriers that maybe we thought they were.”

Techniques like RAG have been reducing hallucinations, and costs per token have been coming down as well, Clarke said. Pairing that with the benefits they get from deploying generative AI, Highmark sees value being created.

As for hurdles to AI agent adoption, Clarke said that AI needs access to data and systems, as well as an orchestration framework for everything to work. Another factor is the organization’s own risk tolerance — how comfortable are they in letting AI agents autonomously complete tasks?

At Highmark, Clarke said they place use cases in “shadow mode” before moving to full automation on tasks in which they have high confidence the AI will do them right, which are typically lower-risk activities.

At Hackensack Meridian, Sethi said the hurdle is deciding which part of a human task they should automate.

“We went through these journeys as a part of the robotics process, automation phase,” Sethi said. “Building the automation, or the agents to actually do this, is not the hardest part. Rather, it’s figuring out what can and should be automated. That’s what slows us down.”

The post Health Systems Tap GenAI to Transform Patient Care appeared first on PYMNTS.com.