Managed Mortality Variance (MMV)
How closely real deaths track the model. A spike is the hardest adverse signal to deny — and the one most likely to break.
Formula
Thresholds & Bands
| Band | Range | State |
|---|---|---|
| On-model | ± 2% | ok |
| Drift | ± 5% | warn |
| Anomalous | > 5% | crit |
Why This Metric Matters
Managed Mortality Variance is the Directorate's most sensitive population-health indicator and its most dangerous exposure vector. When observed deaths diverge from PANACEA's actuarial model, it signals either an uncontrolled adverse effect of fielded compounds or an external variable the model has failed to capture — both of which demand immediate investigation. Mortality spikes are uniquely hazardous because death records are public, independently verifiable, and resistant to narrative reframing. A sustained anomalous reading is the single most likely trigger for external investigation that could penetrate the Directorate's containment architecture.
Threshold Justification
The +/- 2% on-model band represents the natural stochastic variance in population mortality that PANACEA's actuarial models cannot eliminate, accounting for seasonal illness fluctuation, accident rates, and demographic noise. The +/- 5% drift threshold was calibrated to the detection sensitivity of external epidemiological monitoring — variance above this level has historically attracted attention from independent researchers and public-health agencies, making it the operational boundary beyond which containment risk escalates non-linearly.
Historical Context
MMV exceeded the anomalous threshold twice in the first year of monitoring: once during the initial rollout of a broad-spectrum maintenance compound (subsequently reformulated) and once during a regional data-reporting artifact caused by delayed death-certificate processing. Both events prompted significant Directorate response. Since the model recalibration of Q2 2025 — which incorporated compound-interaction effects and regional demographic weighting — the metric has remained within the on-model band for 47 consecutive weeks.
Collection Method
MMV is calculated weekly by R&D Population Analytics using a two-source comparison. Observed mortality is derived from death-certificate registries, hospital discharge records, and VITALNET biometric cessation signals (loss of vital signs from monitored subjects) aggregated through the Synaptic Data Fabric. Expected mortality is generated by PANACEA's actuarial model, which incorporates demographic structure, seasonal baselines, compound-exposure profiles, and known comorbidity distributions. The variance is expressed as the percentage deviation of observed from expected, with directional sign preserved.
Known Failure Modes
Death-certificate reporting lags vary by jurisdiction from 3 to 21 days, meaning that the weekly reading is always a partial snapshot that may be revised upward in subsequent cycles — a pattern that can mask emerging spikes until they are already visible to external observers. VITALNET cessation signals produce false positives when subjects remove or disable monitoring devices, and false negatives when monitored subjects die in areas with degraded mesh connectivity. The actuarial model itself can produce misleading on-model readings if a compound-induced mortality increase is offset by a coincidental decrease in another cause of death, concealing the adverse signal within normal-appearing aggregate numbers.