How one hospital combated ER overcrowding with predictive analytics

A data-driven fix for emergency department crowding

To meet this challenge, Baptist Health collaborated with health tech and operational partner LeanTaaS to launch Operation Raptor, an initiative to optimize patient flow across the facility using a combination of automation and predictive analytics.

"Before this transformation, our morning huddles were driven by spreadsheets and best guesses. Decisions about bed placement and discharges were essentially made in hindsight," said Walker. "Now, our teams come in with a clear, predictive view of who is likely to discharge, when those discharges will happen and where patients will go."

Data sources for analytics-based predictions include the hospital's EHR, staffing systems and external information like weather and flu trends, with updates every few hours. This data feeds into a machine learning model. Staff receive alerts when capacity thresholds are approaching, allowing them to plan discharges, bed turnover and staffing needs proactively.

"We forecast census in 30-minute increments up to eight weeks in advance," said Jason Harber, LeanTaaS' senior vice president of inpatient flow. "We're predicting when a patient will go home, how many beds will be available and even where patients will be discharged to – home, rehab, or skilled nursing." 

Launching a predictive model

For Baptist Health, the most challenging part of launching a predictive analytics program wasn't upgrading technology, it was changing minds.

"Like many hospitals, we had to move from a mindset of 'we know our hospital best' to trusting what the data was telling us," Walker said.

Adoption accelerated when staff validated the data by comparing model-derived and manual predictions to actual discharge times. The AI model predicted discharge times with 90% accuracy versus 60% from manual predictions, according to Walker.

From wait times to measurable wins

Since launching Operation Raptor, the hospital has reduced ambulance diversions by 75%, Walker shared. Average ED wait times have dropped significantly, and staff report smoother handoffs and less pressure during peak hours.

According to Walker, further outcomes include:

• A 34% decrease in inpatient length of stay
• 40% faster discharge processing
• Hundreds fewer patients boarded in the ED each month
• The equivalent of a 30-bed increase in hospital capacity, without any new construction

"What used to take two and a half hours from discharge order to patient departure now averages around 80 minutes," said Walker. "That's helped us create capacity and reduce strain across departments."

Those time savings have also translated into a better experience for both patients and staff.

"Patients spend less time waiting for beds, and staff spend less time firefighting," said Walker. "We've seen noticeable improvements in satisfaction scores and engagement across departments."

Ensuring data accuracy and security

Accurate forecasts depend on fresh data and regular model updates. At Baptist Health, predictions refresh throughout the day to reflect changes in admissions, discharges, staffing and external factors. The predictive model is also continuously monitored for accuracy.

"The system is built to monitor its own performance," said Harber. "If anything drifts off, we catch it and recalibrate."

Along with accuracy, data privacy and security is essential.

"We follow every principle that any cloud computing provider would need to hold sensitive data," said Harber. "Security is foundational in this space."

With success in the emergency department and inpatient flow, Baptist Health is planning to scale Operation Raptor to other areas, including perioperative and ambulatory care.

"The same principles apply across the enterprise," Walker said. "We're looking at incorporating this into repatriation, diversion management and surgical smoothing. The end goal is to build a health system where no patient waits for care, and every decision is informed by data."

Elizabeth Stricker, BSN, RN, comes from a nursing and healthcare leadership background, and covers health technology and leadership trends for B2B audiences.