Build Journal · THC CDSS

How I scope a clinical decision support system for one clinic

By Ganesh R Datta. I build healthcare AI end to end: clinical decision support, clinic intake, and the marketing and operations around them.

The fastest way to stall a CDSS project is to try to support every decision at once.

My scoping questions

• What does the clinic actually deliver? (For Total Health Centers: 12-week heart, brain and longevity programs.)
• Which decision takes physicians the most effort, most often?
• What data is already collected, and where does it live?
• Who will use the tool daily, and who approves its output?

The first version

The first version supports one assessment, for one decision, in the clinic's own program structure. It is in daily use before anything else gets built.

What comes later

Additional assessments, deeper reporting and integrations get added once the first workflow is trusted. Each addition is judged by whether it removes repeated effort from the care team.

Why narrow first

Clinicians judge a tool by its first weeks. A narrow tool that works builds the trust that broader features need.

More from the Build Journal

• What building clinical decision support across streams of medicine taught me
• Decision support, not decision making: where I draw the line in a CDSS
• Mapping AI assessments to the programs patients actually enroll in
• Structuring an integrative assessment around root causes

The full Build Journal · The THC CDSS work · Book a 15-minute call