Researchers from OSF Healthcare and its partners have developed an artificial intelligence algorithm that can predict the upcoming week’s work for each cancer patient navigator for their existing patients. They have also created a second machine learning model that distributes new patients among the navigators within specialties to better balance workloads.

WHY IT MATTERS

Patient navigators can help health systems decrease the time from diagnosis to treatment, but programs are often underfunded, which leads to heavy workloads. 

OSF Healthcare, operated by the Sisters of the Third Order of St. Francis, has 10 acute care hospitals and five critical access hospitals, with 2,084 licensed beds in Illinois and Michigan. The Peoria, Illinois-based health system's cancer patient navigators provide education, advocacy and support to cancer patients and help facilitate their care journeys.

Innovative approaches help to retain employees – "our greatest asset," said Dr. Jonathan Handler, OSF Healthcare senior fellow of innovation, in the health system's announcement.

To achieve greater workload fairness for their pool of CPNs, OSF partnered with the nearby University of Illinois College of Medicine, the University of Illinois at Urbana-Champaign and Northwestern University's Feinberg School of Medicine on a retroactive study to develop and test a machine-learned algorithm's ability to outperform random assignments and create more equitable CPN workloads within a specialty.

The researchers note that one navigator in a specialty might have much more work than others in the same specialty at a given time, they said. 

"CPNs do not transfer their existing patients to other CPNs when workloads become overwhelming," researchers wrote in their report published in the JCO Clinical Cancer Informatics in May. 

"They prefer to retain their patients for the laudable purpose of maintaining a consistent patient-CPN relationship. Therefore, the algorithm uses the only lever available to equalize workloads: the distribution of new patients."

They used a three-year data set compiled from electronic health records, including demographics, cancer types and prior healthcare utilization, to assess the past workloads of 13 specialty CPNs operating at the health system's largest hospital. The data set contained 273,057 records, comprising 13,033 unique patients, according to the report.

The researchers then built three supervised regression models, each one built from one of the most common and successful open-source machine learning libraries. The third step was developing the distribution model that could minimize differences among those navigators in their upcoming week’s workload. 

"Dozens of input features were used to make each prediction each week for every patient," they said.

"Our program seeks to maintain the patient-CPN relationship, so the only consistency constraint imposed was on allocations to ensure that patients remained with their initially assigned CPN throughout their time in the panel."

In addition to their retrospective simulation analysis, the researchers also compared the predictor-informed distribution with a random distribution and assessed resulting workload differences among navigators in the same cancer specialty. They note that anticipated patient needs, navigator experience and existing workload are not considered in OSF's current CPN workload decisions. 

They said that the predictor-informed model achieved significantly greater workload fairness than a random distribution. 

"To our knowledge, this work may represent the first-ever description of an automated, algorithm-driven approach to even out CPN workloads," said the researchers. 

"Optimization has been applied to healthcare staffing and patient allocation in other healthcare domains, but this is usually applied to shifts rather than individuals."

The plan is to integrate the tool OSF Community Connect, a platform that automates workflows, and pilot its efficacy ahead of the planned opening of the OSF Cancer Institute in 2024, according to OSF's announcement last week.

THE LARGER TREND 

Across the globe, AI is being used or developed to address the unprecedented levels of burnout experienced across the healthcare workforce. 

Software companies and healthcare IT developers use machine learning to address a number of healthcare tasks, from transcribing audio or video, addressing administrative inefficiencies and providing insights about patients and patient populations – all to improve efficiencies and patient outcomes and to stop overburdening the professionals that make up the healthcare workforce. 

The UCLA Health System, for example, is using algorithms to make its nursing workloads more equitable. 

Nurse informaticists developed a machine learning model that produced individual real-time workload acuity scores for all nursing staff, said Meg Furukawa and Stesha Selsky, nurse informaticists with UCLA Health. 

Charge nurses use the scores generated, which all nursing staff can see, for decision support, and they can adjust workloads or request additional staff as needed.

Before HIMSS23, they told Healthcare IT News that the ML model relies on existing patient chart information and other nursing documentation from electronic health records and other systems. Furukawa noted that their guiding principles were to create a tool that would not increase administrative burdens and would rely on data generated from existing workflows. 

By working collaboratively, use of AI by UCLA Health's nursing staff has helped to achieve more equitable nursing resources and patient assignments, Furukawa and Selsky said.

"We involved bedside nurses, definitely, from the very beginning, we had nursing leadership, we had our workload acuity champions as part of the project to really give us input and feedback and help us develop the tool and validate the tool along the way," Furukawa explained.

ON THE RECORD

"Our cancer patient nurse navigators are highly dedicated, and their workload can sometimes be overwhelming," said Handler of OSF, in a statement about the new AI findings there. "They never want to shortchange the patient, so they shortchange themselves, working extra hours and sacrificing their own wellbeing to help patients. We hope our system can even out those workloads and improve their work-life balance," he said.

Andrea Fox is senior editor of Healthcare IT News.
Email: afox@himss.org
Healthcare IT News is a HIMSS Media publication.