Every death that enters the registry travels the same five-stage pipeline, rendered here as a real 3D tunnel you can orbit around instead of a flat diagram: Intake (the antemortem clinical diagnosis is logged on the pink rail), Autopsy Examination (the postmortem anatomic diagnosis is logged on the amber rail), Paired Comparison (the two are connected by a beam), Goldman Classification (the beam is graded), and Ledger & Trend (the case is filed and the running concordance trend updates).
P(major) = 0.10 + threshold·0.04 (≈0.14 … 0.22)
P(minor) = 0.14 + threshold·0.03 (≈0.17 … 0.23)
P(concordant) = 1 − P(major) − P(minor)
concordance(n) = 1 − majorCount(n) / n
This is the same classification logic used by real autopsy-discrepancy registries applying the Goldman et al. (1983, NEJM) schema: each case is stochastically graded major (Class I/II, would likely have changed management or survival), minor (Class III, related but clinically unimportant), or concordant. The threshold slider (Lenient → Strict) shifts the adjudication panel's grading toward more or fewer discordant verdicts, exactly as a stricter blinded-adjudication protocol would surface more disagreement. The autopsy rate slider controls how often a new case enters the pipeline, standing in for the fraction of hospital deaths that receive a consented autopsy at all — the registry's entire denominator.
- Autopsy rate — higher rates spawn cases into the pipeline more often; today's real US non-forensic hospital autopsy rate is only ~5-8%, down from ~50% in the 1960s.
- Classification threshold — Lenient/Balanced/Strict shifts the major/minor discordance probabilities, as a real adjudication panel's rigor would.
- Simulation speed — scales how fast cases transit the five gates, so you can slow down to inspect one case's journey or speed up to watch the trend line accumulate.
- Drag to orbit the camera around the pipeline, scroll to zoom, or enable auto-orbit.