When the Kuwait Ministry of Health delegation visited KAMC in December 2025, I had the opportunity to present our nursing informatics work and the administration decision support systems we have built. Here is what we shared and why it resonated.

I have spent the past two years experimenting with AI tools in a real clinical informatics role. Not benchmarks. Not demos. Actual daily use inside a live hospital system. Here is my honest assessment of where AI creates leverage and where it creates noise.

The constraint was simple but hard: every new data source added to the dashboard should require zero changes to existing code. Here is the architecture I landed on using IRIS SQL, DuckDB, and a receiver contract that makes new queries plug in automatically.

At the First Health Management Conference in Makkah, January 2024, I demonstrated the nursing decision support system live at the exhibition floor — including a walkthrough for the CEO of Makkah Health Cluster. The system covers real-time staffing visibility, shortage prediction, and reallocation recommendations.

After 15 years in ICU and informatics, the single biggest frustration I carry is watching good nurses spend a third of their shift on documentation that does not reflect what they actually did — and then watching that data get used to make decisions as if it did.

The official tooling for extracting data from InterSystems IRIS is powerful but heavy. For operational use — quickly pulling a report, checking a value, validating a workflow — you need something portable, low-dependency, and fast. Here is how I approached the design.

I have built dashboards that nobody opened after the first week, and dashboards that became part of daily operations. The difference has almost nothing to do with the visualization library or the color scheme. It is mostly about one thing: does it answer the question the person actually has?