feature: Sepsis Engine Refactor: Remove SIRS, Rewire qSOFA + Bundle

Frontend: GCS Entry, SOFA Display, Sepsis UI Refactor

docs/decisions/sepsis-engine-design.md

@@ -1,4 +1,3 @@
-```markdown
 # Sepsis Engine Design Decisions
 
 ## Why Redis for SIRS state, not a PostgreSQL time-range query
@@ -50,4 +49,21 @@ and no additional infrastructure. The trade-off: if this system needed to scale
 multi-hospital network with 50,000+ concurrent inpatients (~5,000 observations/second),
 Flink would become the right choice. The architecture decision is correct at this scale
 and defensible at interview with a clear scale inflection point named.
-```
+
+## Phase 27 — Migration from SIRS to SOFA
+
+**Rationale:** Sepsis-3 (2016) replaced SIRS with SOFA for organ dysfunction assessment.
+SIRS is non-specific (post-exercise tachycardia, mild fever). Doctor feedback aligned with
+moving sepsis **confirmation** to SOFA delta ≥ 2 while retaining qSOFA as a **bedside screen**.
+
+**What changed:**
+- `SirsDetector` / `SirsEvaluator` deleted — no new `SEPSIS_WARNING` alerts
+- qSOFA ≥ 2 → `QSOFA_SCREEN` (WARNING, suppressible) with lab-order recommendation
+- Sepsis hour-1 bundle triggers from `SOFA_SEPSIS` only (Phase 26 delta ≥ 2)
+- Historical `SEPSIS_WARNING` and `QSOFA_WARNING` rows remain queryable
+
+**Redis key patterns after Phase 27:**
+- `qsofa:{encounterId}:{code}` — qSOFA screening (30 min TTL)
+- `sofa:{encounterId}:{code}` — SOFA lab carry-forward (Phase 26)
+- `gcs:{encounterId}:{code}` — GCS components (Phase 25)
+- ~~`sirs:{encounterId}:{code}`~~ — removed (legacy keys expire)