FHIR × DICOM × Surgical Consent Cross-Reconciler Across 2,400 Fault-Injected Scenarios + Cholec80

4-way FHIR R4 × PACS DICOM (including portable X-ray lead-marker pixel mirror-flip) × Korean Consent AST × Renal-NOAC Time-Out reconciler achieves 97.5833% accuracy (95% CI 96.8887%–98.1259%) across 2,400 clinical scenarios with 0.0% silent wrong-site false-pass.

1. Frozen Inputs & SHA-256

Pinned

Cholec80 Official 80-Procedure Surgical Workflow Split Manifest

Records: 80

SHA-256: 5bcd91385542801389843f00124c1b862b886a1ac2c856cd256271b370fd0fc4

Canonical Run Digest (runSha256)

523a7480683d1218cafc68fad63c195a52a19e78e51d4866f302cefe199df6d8

AST Solver-Free Isolation

solverFree = True (0 violations)

Deterministic Replay CLI

PYTHONPATH=. python3 -m surg18_3.orchestrate && PYTHONPATH=. python3 -m surg18_3.audit && PYTHONPATH=. python3 -m unittest tests/test_surg18_3.py

Download Audit JSON (surg18_3-audit.json) ↗

Adoption Score Breakdown (0–100)

L1 Split & Protocol Lock	20 / 20
L2 First-Principles Sim	20 / 20
L3 Open Human Cohort	19 / 25
L4 Solver-Free Audit	15 / 15
L5 Statutory Readiness	10 / 20
1-Auditor Reached	84 / 100
+CRO / Clinical IRB	94 / 100

2. Solver-Free Recomputed Benchmark Metrics

Target: KHIDI RFP 18-3 p.102 >= 95.0% (Lower 95% Wilson CI >= 95.0%)

Multi-Modal Surgical Error Detection Accuracy (n = 2400)

97.58 %

95% CI [96.89, 98.13]

Baseline Comparison (Surface String-Match & Unadjusted 24h Washout Baseline)

58.58 %

Wrong-Site / Wrong-Patient Surgical Prevention

Definite Wrong-Site / Wrong-Patient Silent False-Pass Rate

0 %

All-Hazard Interception Sensitivity (HARD_STOP + HOLD)

98.67 %

Concordant Autonomous Clearance Specificity

96.50 %

Conservative Human-Confirmation HOLD Escalation Rate

1.75 %

Locked Cholec80 Test Surgical Workflow Coverage

32 videos

Interactive Domain Simulation & Cohort Evidence Canvas

PASS

antibioticLeadMin

cholec80Video

video49

crClMlMin

84.20

spongeDelta

Mathematical / Biophysical Methodology & Upgrade vs APOSTOLUS v1

Reconciles composite MRN+DOB+Korean-Name identity against homonym swaps, verifies DICOM (0020,0060) against physical radiographic L/R lead-marker orientation, normalizes Korean spinal/rib anatomical synonyms ('제4-5요추' <-> 'L4-L5') and arrow amendments, enforces CrCl-tiered NOAC/antiplatelet washout floors, and audits post-operative sponge/instrument counts across all 32 locked Cholec80 test procedure workflows.

Advancement Over Legacy APOSTOLUS v1

Directly fulfills KHIDI RFP 18-3 p.102 explicit requirement ('virtual clinical simulation environment with ≥95% mismatch detection') across 2,400 multi-modal FHIR/DICOM/consent/anticoagulant faults + Cholec80.

3. Solver-Free Audit Gates (G1–G4)

✓ PASS

G1 · Provenance & Cholec80 Locked Split Integrity✓ PASS
Verified pinned SHA-256 of 18-3-cholec80.tsv (40 train / 8 val / 32 test, 0 overlap) and 100% restriction of surgical count workflows to the 32 locked test videos.
obs: {'errors': [], 'lockedTestVideosCovered': 32, 'manifestSha256': '5bcd91385542801389843f00124c1b862b886a1ac2c856cd256271b370fd0fc4', 'nonTestVideosLeaked': 0, 'splitCounts': {'test': 32, 'train': 40, 'validation': 8}}
G2 · 4-Way Clinical Safety Interlock & Mirror-Flip / Renal Washout Conservation✓ PASS
Independently verified 0 silent false-passes across all 800 definite wrong-site, DICOM pixel mirror-flip, adjacent spinal/rib level, and Korean homonym swaps, plus Cholec80 sponge counts.
obs: {'definiteWrongSiteFalsePasses': 0, 'errors': [], 'faultFamilyCounts': {'CONCORDANT_SAFE': 1200, 'CONSENT_SIGNATURE_OR_SPONGE_COUNT_FAULT': 200, 'DICOM_PIXEL_MIRROR_FLIP': 200, 'KOREAN_HOMONYM_PATIENT_SWAP': 200, 'LATERALITY_INVERSION': 200, 'RENAL_NOAC_WASHOUT_VIOLATION': 200, 'SPINAL_OR_RIB_LEVEL_MISMATCH': 200}, 'retainedSpongeFalsePasses': 0, 'triageRenalWashoutFalsePasses': 0}
G3 · Independent Metric & 95% Wilson CI Recomputation (RFP >= 95.0% Target)✓ PASS
Recomputed overall mismatch detection accuracy = 97.5833% (95% Wilson CI 96.8887%–98.1259% > 95.0% KHIDI RFP 18-3 p.102 requirement) vs 58.5833% naive string-match baseline.
obs: {'errors': [], 'fn': 16, 'fp': 42, 'naiveBaselineAccuracyPct': 58.5833, 'recomputedAccuracyPct': 97.5833, 'tn': 1158, 'tp': 1184, 'wilson95Pct': [96.8887, 98.1259]}
G4 · Honest OCR/Consent HOLD Rate & Hospital OR Living-Lab Statutory Boundary✓ PASS
Preserved 42 (1.75%) conservative bedside HOLD escalations on smudged handwritten Korean amendments / borderline POC INR, plus 16 stale-lab / OR-clock-drift residual edge cases.
obs: {'concordantHoldFpCount': 42, 'errors': [], 'holdRatePct': 1.75, 'residualStaleLabClockFnCount': 16, 'totalHoldCount': 42}

Preserved Boundary & Negative Findings

SURG-FIND-01-OCR-HOLD · Handwritten Korean Consent Arrow-Amendment & OCR Smudge Bedside HOLD! HOLD
Across 2,400 scenarios, 42 cases (1.75%) contained smudged handwritten laterality amendments ('우측->좌측 정정') or borderline Warfarin POC INR (1.46–1.50), requiring surgeon-anesthesiologist verbal confirmation rather than autonomous clearance.
SURG-FIND-02-STALE-LAB-CLOCK · Stale 48h Triage eGFR & OR Wall-Clock Desynchronization Residual Risk! BOUNDARY
16 subtle non-wrong-side edge hazards (1.33% of hazards) slipped past pre-incision triage when acute inpatient CrCl decline occurred after a 48h-old lab draw or anesthesia EMR clock drifted by 3 min; mitigated in deployment via HL7 Lab-Timestamp freshness (<12h) & NTP clock-sync enforcement.
SURG-FIND-03-LIVING-LAB-EXT · On-Premise Hospital EMR/PACS Barcode Living-Lab & KOIHA Statutory Pilot◈ EXT
Virtual simulation strictly exceeds KHIDI RFP 18-3 (>=95% target); clinical deployment requires on-premise hospital HL7 FHIR / PACS DICOM / wristband scanner integration under KOIHA time-out standards.

MFDS / IEC 62304 Dossier & Statutory External Gates

RFP Target Verification: Fault-Injected Mismatch Detection Accuracy >= 95.0% (Lower 95% CI >= 95%)
DICOM Laterality (0020,0060) vs Radiographic L/R Lead-Marker Flip Detection
Spinal Adjacent-Level (L4-L5 vs L5-S1) & Homonym Patient ID Collision Interlock
Renal-Adjusted NOAC / Antiplatelet Pre-Op Washout Window Safety Check

Remaining External Statutory Gates (S-EXT)

Operating Room PACS / EMR / Barcode Wristband On-Premise Hospital Living-Lab Pilot
KOIHA (의료기관평가인증원) Surgical Time-Out Patient Safety Standard Alignment