Federal Health Agencies Are Using AI to Improve Care

Federal health agencies are deploying artificial intelligence to improve patient care, using the technology to help prevent nursing home injuries, expand clinical training and connect pediatric health data across institutions.

Officials from the Health Resources and Services Administration, Department of Veterans Affairs and Advanced Research Projects Agency for Health highlighted how AI is moving into clinical and workforce applications Wednesday at GovCIO Media & Research’s Health IT Summit.

Better Nursing Home Care with AI

The Health Resources and Services Administration is incorporating AI into its Telehealth Centers of Excellence programs, which implement and evaluate innovative telehealth approaches for rural communities and areas with limited access to care, HRSA Administrator Tom Engels said Wednesday.

Engels cited the University of Mississippi Medical Center, which is testing an AI-enhanced virtual nursing home with HRSA funding. The model uses camera-based software to monitor certain patient movements and alert nursing staff when it detects a potential fall risk.

“The AI does not replace the nurse; it does not decide what the patient needs. It identifies a possible risk and alerts the staff to a possible danger for that patient,” Engels said.

HRSA also supports the National Telehealth Technology Assessment Resource Center, which provides technology assessments, technical assistance and resources to help health care organizations determine which technologies best fit their needs.

Engels recently attended a demonstration of emerging technologies developed at the center. One promising effort uses a tablet camera to capture a short video of a patient’s face and analyze the footage to estimate blood pressure and other vital signs. Other technologies include a virtual stethoscope that connects to a tablet or smartphone.

“What stood out to me is the value of seeing these tools in action and asking questions directly to the technology experts. An organization needs to understand what technology can and cannot do, and determine what best addresses the needs of investing and what they best need for their communities, we want to build in this work,” Engels said.

Virtual Clinical Training at the VA

The Las Vegas VA Medical Center has developed an AI-assisted virtual training system that allows clinicians and other staff to complete training around their schedules, said Bonnie Barti, a simulation technician at the center.

“We needed a way to deliver realistic class of training without interrupting patient care,” Barti said.

The system is designed to serve off-campus, night-shift and weekend staff, as well as other personnel with alternative schedules. It combines two training modes on a single platform: AI-generated virtual patients and a 360-degree immersive virtual environment.

The 360-degree environment simulates situations in real-world workspaces, such as triage during a mass-casualty event. Staff can access the training from a computer or through a virtual reality headset.

A key feature of the platform is that it gives VA clinical staff control over its content. During development, the team prioritized retaining in-house authoring capabilities so staff could create training tailored to their needs.

“It gives us freedom to build our own scenarios directly relevant to our VA healthcare,” she said.

Using AI to Support Pediatric Care

The Advanced Research Projects Agency for Health (ARPA-H) is working to make pediatric care more efficient by using AI and interoperable data systems to give clinicians better access to patient information.

ARPA-H’s Pediatric Care eXpansion (PCX) program aims to connect more than 200 pediatric hospitals and care centers through a national interoperability and analytics layer, said Erika Kim, program manager for resilient systems at the agency.

In its initial phase, PCX supports four areas of complex pediatric care: pediatric brain tumors, epilepsy, congenital heart defects and other rare diseases affecting children.

The goal is to provide clinicians with a more complete picture of a patient’s health. Kim said health data currently resides in numerous silos, limiting its effective use and contributing to delayed diagnoses and treatments.

That fragmented structure can also place the burden on families to find and coordinate the best care for their children. PCX aims to shift that burden “from families to the infrastructure itself, so that relevant clinical, imaging, pathology, laboratory, and genomic information can securely follow the child across participating care settings,” Kim said.

PCX enables standards-based information exchange across electronic health records, laboratory systems, pathology, imaging and genomics. AI and natural language processing can structure difficult-to-use clinical information, including narrative notes and pathology reports, while supporting privacy-preserving access and more timely analysis.

The program is also exploring federated approaches that allow AI models to learn across institutions without requiring organizations to move patient data, Kim said.

Another key component is developing AI capabilities for tasks such as matching pediatric cancer patients with clinical trials. As more hospitals connect to the system, clinicians could also use collective data to understand how patients with similar conditions responded to different treatments and use those insights to better inform clinical decisions.