• Article
  • November 11, 2020

Using Predictive Analytics to Assist Mental Health Providers

Seattle Children's developed a program to inform providers of the admission decision making process.

As a regional mental health facility for children under 12, Seattle Children's often faced tight capacity constraints on its Psychiatric-Behavioral Medical Unit (PBMU). With only 41 beds, and a high number of internal and external placement requests, the unit is often full. This, combined with the complexity of mental health recovery and length of stays, led hospital leadership to take action to improve patient access.

Ruth McDonald, M.D., chief medical officer, and T. Eugene Day, data science manager at Seattle Children's, presented at the 2020 Annual Leadership Conference how the hospital's analytics team set out to create a predictive analytics program to monitor and predict the capacity of the PBMU.

The program, FlowAdvisor, provided the opportunity to serve the community, improve care in the hospital and recapture lost revenue from over-buffering.
"We're often making tradeoffs between system performance and revenue, or speed and quality. However, many times you can improve both with the appropriate use of predictive and prescriptive analytics," Day says.

Right child, right bed, right time

FlowAdvisor was built to provide real-time data based on location, length of stay and patient census cohorts. This provided the unit with information regarding who was in the PBMU, whether they had likely discharge decisions, and patients in other units who had mental health diagnoses.

"All of this information was already available," Day says. "But it was just scattered across the EMR and required a lot of searching."

To determine the probability of emergency department arrivals and other predicted information at any given time, Seattle Children's used historical demand based on the day of the week, time of year and whether school is in session.

This information is meant to assist providers with the decision-making process, not replace it, according to Day and McDonald.

The adjustment period

Throughout the implementation process, weekly meetings with PBMU leadership and staff allowed everyone to learn and grow together.

"We wanted the app to be useful at each stage in the process of development," Day says.

To develop FlowAdvisor to best fit the unit's needs, the analytics team at Seattle Children's started small and added or improved capabilities as it was relevant to the providers.

While the team made progress with FlowAdvisor, the hospital paused the program at the onset of the COVID-19 pandemic.

Send questions and comments.