I’ve been developing BirthSense OS on top of OpenMRS and working with clinicians who highlighted documentation challenges with current AI scribes: they add irrelevant information to notes, miss critical details, and create errors in roughly 1 out of 20 notes, with no way to trace back to source material.
Built BScribe to solve the source grounding problem. Every extraction maps to its exact location in the original clinical text, allowing clinicians to instantly verify where information originated through visual highlighting.
We’re implementing direct integration with OpenMRS to generate clinical notes with complete source traceability, retrieve relevant clinical guidelines and CDC recommendations with verification links, and maintain compatibility with existing OpenMRS workflows. The goal is to leverage OpenMRS’s modular architecture while adding AI documentation that clinicians can actually trust and verify.
Looking for feedback from the OpenMRS community on technical implementation approaches, workflow patterns that would be most valuable to automate, and ideas for deeper integration. Has anyone else worked on similar AI documentation tools with OpenMRS? Would love to hear about your experiences.