Enterprise

Parkland Hospital Uses AI to Predict Patient Violence Risk

Dallas scientists developed a predictive model that identifies high-risk patients within 15 minutes of admission, addressing a nationwide crisis in healthcare worker safety.

Omega Editorial · August 23, 2026 · 3 min read

Key takeaways

  • Parkland Health's AI model predicts patient violence risk within 15 minutes of admission, with 77% of high-risk patients involved in violent incidents
  • The system analyzes factors like arrival method, infections, medications, smoking history, and neighborhood vulnerability data without considering race or ethnicity
  • Eight out of 10 nurses experienced workplace violence in 2024, driving retention problems and labor actions across the healthcare industry
  • PCCI is expanding the model to Northwestern Memorial Hospital in Chicago and partnering with Vega Health for wider deployment
  • The tool flags roughly 30% of patients as higher risk, allowing staff to focus protective resources where they're most needed

AI model targets healthcare violence epidemic

Parkland Health has deployed an artificial intelligence system that predicts which patients pose a violence risk to hospital staff within 15 minutes of admission. The Workplace Safety Model, developed by the Parkland Center for Clinical Innovation, represents a shift from reactive security measures to proactive risk assessment at a time when healthcare worker safety has reached crisis levels.

The tool's validation data shows that among patients flagged as high risk, 77% were involved in violent incidents, according to PCCI staff. The system allows Parkland to focus resources on approximately 30% of patients identified as higher risk while clearing the majority who pose no threat.

How the prediction system works

The AI model analyzes multiple factors without considering race, ethnicity, or religion. Each patient receives a fresh risk evaluation at every visit rather than carrying a permanent profile.

Key variables include whether the patient arrived by ambulance, presence of severe infection, current medications, and smoking history. The last factor matters because Parkland maintains a tobacco-free campus, and nicotine withdrawal can increase irritability.

The system also draws on PCCI's Community Vulnerability Compass, which maps social and health vulnerabilities by census tract. A patient's address provides insight into life stressors such as financial instability or food insecurity that might correlate with behavioral risk.

"Every time a person comes in, [he or she] gets a new risk evaluation," said Steve Miff, PCCI president and CEO. "We build this in a way that it predicts an instance, not a person."

Jacqueline Naeem, vice president of clinical and social health at PCCI, emphasized the tool supports rather than replaces clinical judgment. Staff use the assessment to determine whether additional precautions are warranted.

Why it matters

Healthcare workplace violence has accelerated dramatically. In 2024, 8 out of 10 nurses reported experiencing violence at work within the preceding year. The problem has driven staff retention issues and even sparked strikes in some states, with workers reporting being shoved, pushed, scratched, and bitten by patients. Traditional hospital responses—personal alarms and video surveillance—only react after incidents occur. Parkland's predictive approach could prevent violence before it happens, protecting both staff and patients while reducing the substantial costs hospitals bear from these incidents.

Expansion beyond Dallas

PCCI is preparing to implement the model at Northwestern Memorial Hospital in Chicago and has partnered with healthcare technology company Vega Health for broader deployment. The expansion suggests growing recognition that predictive tools may offer a more effective approach than conventional security measures alone.

The details were first reported by The Dallas Morning News editorial board.

  • #healthcare ai
  • #workplace violence
  • #predictive analytics
  • #hospital safety
  • #parkland health
  • #clinical innovation

This is an original analysis by the Omega editorial team. Source reporting: AI Watch.

Want systems like this working for your business?

Book a Call

More in Enterprise