SafeMedAI is a domain-driven healthcare compliance platform — combining blameless error reporting, AI-powered root cause analysis, anonymous patient feedback, and continuity of care into a single, audit-proof system built for O'zbekiston tibbiyoti.
Core Architecture
Every design decision traces back to a single constraint: medical systems demand correctness before convenience.
Pillar 01
Domain-Driven Design — each app is a bounded context. Business logic never bleeds into views. HTTP is just transport.
Pillar 02
Pillar 03
Pillar 04
Four independent, pluggable engines — each with a stable typed interface. Swap keyword-based sentiment for an ML model tomorrow without touching a single consumer.
Pillar 05
API-first from day one. OpenAPI 3.0 auto-generated. Frontend, mobile app, or third-party system can connect without backend coordination.
Root Cause Analysis
Anonymous Patient Feedback
Keyword scoring across three languages. Rating carries 60% weight; text carries 40%. Returns a typed result with a confidence score.
Bridges the discharge-to-home gap. Auto-assigns the least-loaded family doctor by region. Tracks vital signs per visit. No patient falls through the system.
DB-backed, Redis-free. 7 typed alert categories with delivery tracking, retry logic, and per-message error capture.
Compliance Core
11 action types. Field-level JSON diff on every mutation. IP context
on every write. No one — including system admins — can delete an audit record.
save() raises an error
on existing records. Append-only by design.
Ready to Deploy
Domain-driven. Audit-proof. WHO/JCI aligned. Built for Uzbekistan's healthcare system — deployable anywhere PostgreSQL runs.