University of Pennsylvania

01/15/2025 | Press release | Distributed by Public on 01/15/2025 11:10

The future of nursing care

Even before an abnormal lab result comes back, nurses pick up on subtle, troubling cues and adjust care. In thousands of medical records, nurses leave a trail of warning signs noticed and addressed. This wealth of life-saving information has long been locked away in dense, disparate nursing notes-but now there is a key.

Illustration: Mary Haasdyk Vooys

With the power of artificial intelligence (AI), Kenrick Cato, Penn Nursing faculty member and professor of informatics at the Children's Hospital of Philadelphia, co-created a clinical decision support early warning system that accurately detects patient deterioration.

By modeling nursing concern and expertise, CONCERN (Communicating Narrative Concerns Entered by Registered Nurses) seeks to reduce the hundreds of thousands of annual inpatient deaths from cardiac arrest and sepsis by prompting providers to timely intervention.

Cato and his co-developers started with natural language processing (NLP): Think of a computer brain scanning and making sense of thousands of nursing notes. With the resulting data, they developed machine learning (ML) algorithms that enable CONCERN to quickly assess 200-plus patient variables and assign a risk level (low, medium, or high). A health care team can view the risk assessment on computer, tablet, or mobile phone. Easily integrated into all major electronic health record platforms, CONCERN is, says Cato, "one of the first AI-based clinical decision support systems to be rigorously evaluated-and the first to use nursing data and be nursing-led."

AI-based systems automatically coalesce health record information across transitions in care to ensure important details don't go astray, and even predict which patients require more attentive planning for positive outcomes. Machine learning algorithms predict heart failure, prompting clinicians to intervene earlier. Nurse researchers, too, grapple with large amounts of data with the help of AI. "With machine learning and natural language processing, you can extract meaning from a large body of text, images or other materials, and gain insights much more quickly," says associate dean for Research and Innovation and Penn Integrates Knowledge University Professor George Demiris.

As he works to ensure Penn Nursing is a thought leader and convener in the AI and health care space, Demiris is both optimistic and cautious about this technological revolution.

"We want to be at the forefront. As we think about AI in nursing, we need to decide or at least recommend what the path should be, rather than waiting for the technology development to dictate how things are going to be," he says, adding that the School brings nursing expertise to the conversation, along with ethical, legal, and policy implications.

To that end, faculty are using AI in their research and leading the development and implementation of AIHTs to improve the lives of patients and nurses. The School is creating courses that prepare students to think critically about AI tools; hosting events and workshops; and forging interdisciplinary partnerships, on campus and across the United States, with the aim of designing useful and ethical solutions.

Where other early warning systems that rely on physiological data alone predict deterioration 12 hours before an event, CONCERN reliably spots risk signs up to 72 hours prior. According to Cato, clinical trials at two hospitals found the system significantly decreased average length of stay, mortality, and sepsis. There was a significant increase in unanticipated transfers to ICU and patients being discharged alive.

With a promise to improve patient outcomes as well as workflow, the release of artificial intelligence health care technologies like CONCERN is rapidly increasing. As of May 2024, the FDA has authorized nearly 900 AI/ML-enabled medical devices. Tech companies, from startups to giants like Google, are investing heavily in the AI health care space.

This story is by Janine White. Read more at Penn Nursing News.