Published

0 min read

How AI helps teams act on recurring risk

How AI helps teams act on recurring risk Engaged and Disengaged Providers at Conference Table GettyImages 1149286425

By Mike Personett, Senior Vice President, Safety and Reliability Solutions

Resolving a safety event doesn’t always resolve the risk behind it. The same problem can return under a different name, in a different unit, or at a different facility. Each event may be reviewed and addressed, yet the conditions connecting them persist. For safety leaders, the challenge is to recognize when separate incidents are evidence of a larger problem, then intervene before it causes more harm.

AI provides a way to make those connections clearer. By helping teams examine narratives and findings across events, it can reveal recurring risks that typically are difficult to spot through individual review. The opportunity goes beyond faster analysis: It’s also to turn collective learnings into changes that address the underlying conditions—and verify that those changes make care safer.

In safety, improvement often begins with a single event

Teams analyze what happened, identify contributing factors, and implement actions intended to prevent recurrence. That work is essential, but a single-event lens can be limiting. When organizations focus only on the event in front of them, they can miss broader patterns across units, locations, workflows, and patient populations.

The result is familiar to many safety leaders: déjà vu events that recur because the underlying system issue was never fully addressed.

Breaking that cycle requires connecting what teams learn across events. AI can support that work by helping to identify shared risks, target the conditions behind them, and track whether interventions make care safer.

1. Connect related events and shared drivers

AI can analyze event narratives, and review findings as well as other unstructured data to help safety teams see when events that appear unrelated share a common system driver. A medication delay, a missed escalation, and a documentation gap might sit in different categories, but together they may point to a workflow issue during handoffs or shift changes that requires system-level attention.

Once related events are identified, AI can help summarize recurring contributing factors, such as communication breakdowns, handoff issues, documentation gaps, staffing constraints, workflow variation, and technology-related friction. But this doesn’t make AI the decision-maker. Instead, it gives safety teams a clearer way to see relationships that aren’t visible through manual review alone. As a result, teams can move from asking only, “What happened in this event?” to asking, “Where else is this pattern showing up, and what system conditions are allowing it to continue?”

2. Design actions that address the recurring condition

Understanding patterns helps safety teams target the conditions that allow the same problem to recur, rather than focusing solely on the individual event. If, for example, a pattern points to handoff variation, the response may be to standardize required information at shift change or embed a structured handoff step into the workflow. If events cluster around delayed escalation, teams may need to clarify escalation criteria, adjust monitoring triggers, or define who owns the next action when a patient’s condition changes. If documentation gaps appear repeatedly, the better intervention may be to reduce duplicative fields, prompt for missing information at the right point in the workflow, or redesign the process so documentation supports care without creating more work for caregivers.

3. Monitor interventions to see if they’re working

AI also helps support ongoing monitoring after improvement actions are implemented. By tracking recurrence, related signals, and post-intervention trends, organizations can better understand whether an action reduced risk or whether additional changes are needed.

This is where prevention becomes measurable. Improvement work shouldn’t stop when an action plan is documented. Safety teams need to know whether the risk level actually changed.

Trust, governance, and adoption determine value

AI changes safety workflows by reducing the manual effort required for tasks that have historically been difficult to scale. The efficiency opportunity is real, but the larger value comes from preventing harm and improving care processes.

To realize that value, leaders need to prepare safety teams for new ways of working. That includes training people to interpret AI outputs, validate recommendations, monitor for bias, and use AI-generated insights as inputs into human-led learning and improvement, with clinical expertise and local context guiding final decisions.

Trust is essential. Organizations need transparency about how AI is used, safeguards against punitive applications, and clear governance over data, models, workflows, and accountability. AI should be positioned as a learning tool rather than a surveillance mechanism. If caregivers see AI as a way to reduce burden, strengthen follow-up, and prevent harm, it can reinforce psychological safety and support a culture focused on learning.

Moving from reporting to prevention

The goal isn’t simply to detect more events or automate more tasks; it’s to evolve from reporting systems into learning systems—systems that see risk more clearly, learn more deeply from harm and near misses, and act faster on the patterns that contribute to preventable harm.

Used responsibly, AI can help organizations connect event-level learning to system-level action, measure whether recurring risk is declining, and move closer to the goal of highly reliable care.

This is the final post in a three-part series on how artificial intelligence can strengthen healthcare safety. Miss an article? Start from the beginning. Read part 1 here >>