Integrating AI with patient safety data to reduce harm
Duration: February – October
Role: UX/UI Designer
CONTEXT
Patient safety and harm events
The Quality Team oversees the collection, analysis, and reporting of patient safety incidents across The Fraser Health Authority. Patient safety reports enable front-line staff to document any event that caused harm or had the potential to cause harm to a patient. This may be anything from an unsupervised fall to a delay in an individual's lab results.
The current tools allow the Quality team to view reporting data, pull specific reports, and identify areas for improvement, but the process remains highly manual. This gap created an opportunity to design a new solution that better organizes data, reduces manual effort, and provides clearer visibility into key areas of patient harm.
PROBLEM
Valuable lower-harm incidents are being overlooked
- The existing PSLS interface is outdated and requires significant manual work.
- Most incidents fall under Harm Levels 1–3 but are deprioritized in favour of severe cases.
- Teams often respond reactively instead of identifying early warning signs.
- Report creation is slow, inconsistent, and requires manual theme extraction.
SOLUTION
An AI-data-driven patient safety platform
- AI-theming engine to categorize large volumes of safety reports.
- Instant access to low-harm and high-harm event metrics used to flag key areas.
- A modern and user-friendly dashboard for exploring safety trends and themes.
RESEARCH & INSIGHTS
Understanding workflows, pain points, and data patterns
Through stakeholder interviews, workflow observations, and iterative feedback, several key insights emerged:
- Coordinators spend three weeks manually reviewing and generating monthly reports for each of the sites.
- Patterns across lower-harm events may go unnoticed until resources are available to address them.
- Site patient safety leadership committees need consistent and timely reporting packages.
- The legacy interface slows down analysis and does not support proactive action.
How Might We...
IDEATION
Exploring modern solutions for complex data
Early concepts focused on:
- Dashboard layouts that improve data visibility.
- Clear visualizations for incident trends and harm levels.
- System for flagging areas with potential for high-harm events to occur.
- Guided workflows for report creation.
FINAL DESIGN
A modern AI-supported safety platform
The final solution includes:
- AI-supported themes and sub-themes compiled from PSLS report data.
- Data-driven dashboard for exploring different key harm areas.
- Risk flagging for proactive intervention.
Projected Results
KEY TAKEAWAYS
Reflection
Designing a patient safety platform demonstrated how approaching the problem of patient harm proactively might shift organizational culture, helping teams move from reacting to harm events toward preventing them. This project reinforces trust in AI algorithms, not as a means to replace healthcare staff but to support their work and improve efficiency, leading to better quality care across the health authority.
