Psychiatric Care AR
Designing augmented reality interactions to make psychiatric care safe
Pulse uses low power radar technology to protect people at the most vulnerable time in their lives. It maps high risk areas in psychiatric care wards so that staff can be alerted when a dangerous situation occurs. This enables intervention reducing the instance of unattended medical emergencies, self harm and suicided. I led design development with Safehinge Primera in preparation for market deployment and product validation. This included primary research, product roadmapping and value proposition development, developing UX and design systems and prototyping commercial-ready WCAG compliant interfaces.
Product Roadmap
UI & UX Design
WCAG Compliance
Design System
Respectful Patient
Safety Monitoring
Unlike camera-based systems, radar offers a balanced approach to delivering critical location and safety information without compromising privacy. The interface intentionally limits occupancy detail, using colour cues and alert triage to focus care workers' attention on areas that require immediate support.
The design was shaped by auditing existing proof-of-concept systems, creating a new triage framework to handle various alert types, and prototyping multiple concepts for evaluation. A final system was then developed, incorporating feedback from testing.
Response Coordination
& System Readiness
To improve alert visibility, a ward office portal was developed. It displays alert locations within the ward layout, helping staff coordinate responses more effectively. The system also monitors active staff handsets to ensure teams on the ground maintain continuous data access so they can respond efffectively.
In building the portal, existing customer ward plans were analysed, and a range of visualisation concepts was developed. These were assessed for utility, and the final framework was pressure-tested against all known configurations to ensure robustness and scalability.
Algorithm Training
& System Improvement
To continuously improve algorithm accuracy and reliability, a training mechanism was built into the interface. This allowed staff to quickly categorise inaccurate alerts at the point of dismissal, feeding data back into the system for ongoing refinement and model training.
Developing this mechanism involved close collaboration with the data science team to map all conceivable scenarios that could produce inaccurate data. This was consolidated into a feedback information architecture designed to preserve response times for critical alerts.