ICU Information Management System
Live bedside ICU platform — vitals, charting, early-warning prognosis.
Highlights
- Built the clinical frontend for real-time ICU monitoring and charting
- Designed the relational data model with a code-first approach and optimized query performance for bedside page loads
- Integrated live medical-device telemetry streams into bedside patient timelines
- Surfaced early prognosis alerts on the bedside UI for faster clinical intervention
- Authored templated PDF epicrisis reports for clinician sign-off and record export
- Rebuilt the platform from a discontinued first version, migrating clinical workflows into the new architecture
Problem
Three pressures stacked. The previous version was being discontinued and aging out, so years of clinical workflow knowledge had to migrate into a clean architecture without losing what nurses and intensivists already knew how to do. Bedside performance was a daily complaint — charting screens needed sub-second loads while showing live vitals alongside a working chart. And the early-warning loop was broken: prognosis signals existed but lived in a separate analytics tool the bedside team rarely opened.
Approach
Built the clinical frontend in React/TypeScript on a .NET backend with a code-first PostgreSQL data model — relational on purpose, with query patterns tuned to the bedside page rather than to abstract reporting. Live medical-device telemetry streams into the patient timeline through HL7-aligned device feeds and vendor-specific bridges, so a vital sign and the nurse’s note land in the same place. Early-warning prognosis surfaces directly on the bedside UI instead of behind a tab. Templated PDF epicrisis reports come out of Puppeteer + Handlebars for clinician sign-off and record export.
Result
The platform ran in production across all 50 US hospitals and 100+ hospitals in Turkey, replacing the discontinued first version as the primary ICU charting system at every one of those sites.