-
Notifications
You must be signed in to change notification settings - Fork 0
Age 70 Mortality Analysis Results
This page details the survival outcomes for the Age 70 cohort (N=136,530) over a 703-day follow-up period. We employ a triangulation of three distinct statistical philosophies to ensure the survival signal is robust, valid, and logically sound.
The following table summarizes the Life-Days Gained (ΔRMST) per person across all experimental scenarios.
| Analytical Framework | Philosophy | Real Data (FOI) | Null (HR=1) | Stress (5% Error) |
|---|---|---|---|---|
| Hernán Methode | Causal (Trial Emulation) | +17.32 Days | -0.48 Days | +6.27 Days |
| Empirical (Historical) | Biased History (Raw) | +15.10 Days | -1.23 Days | -17.47 Days |
| Empirical (CC) | Adjusted Reality (Matched) | +14.80 Days | -1.80 Days | -20.10 Days |
| C.S. Peirce | Weighted Logic (Evidence) | +2.49 Days | +0,324 Days | -0.747 Days |
This framework emulates a randomized trial by cloning, censoring, and weighting to estimate the biological potential of the vaccination strategy.
- Result: 17.32 Days gained per person.
- Vaccine Effectiveness (VE): +62.9%.
- NNT (Number Needed to Treat): 21.07 to save one full life-year.
- Stress Sensitivity: Dropped to 6.27 days when 5% of data was misclassified.
This compares the raw historical record against a version adjusted for Immortal Time Bias using dynamic case-control matching.
- Historical (Biased): Measured at +15.10 days.
- Case-Control (Adjusted): Measured at +14.80 days.
- Bias Gap: The difference of 0.3 days represents the magnitude of Immortal Time Bias within the Czech registry for this specific cohort.
This framework uses Information Theory to separate "Signal" from "Noise" based on the "Weight of Evidence."
- Weighted Core (Real Data): 2.49 Days. This represents the survival benefit that is mathematically irrefutable even under conservative informational weighting.
- Abduction Diagnostics: Identified 395 "High Surprise" days where the survival advantage logically required a non-random explanation.
- Stress Resilience: In the 5% Stress Test, the core evidence collapsed to -0.747 Days, correctly flagging the data as corrupted/noisy.
To ensure the pipeline does not manufacture results, we ran all engines against a simulated dataset where the vaccine has no effect.
- Verdict: All methods correctly converged to Zero (range: -1.8 to +0.3). This confirms that the results observed in the Real Data are genuine signals.
We intentionally moved 5% of unvaccinated records into the vaccinated group to simulate registry errors.
- Observation: The survival signal was preserved but significantly dampened across all methods.
- Conclusion: The analysis is highly sensitive to data integrity, proving that the high survival gains found in the Real Data are dependent on the high quality of the primary records.
- Causal Analysis: AC) hernan_style_poold_logistics_RMST
- Empirical Analysis: AE) emperical_dynamic_CC_RMST
- Peircean Analysis: AE) C.S. Pierce evidence weighted RMST
-
Bias-Necessity RMST Audit: Methodological Framework Bias Necessity via Biology only Falsification
-
Full Run Results : Full Run Results – Bias‐Necessity RMST Audit
- Descriptive Event-Time Hazard Study : Run Results - Mortality Timing Around Vaccination Eligibility in Czechia
- AG70 results and data files: Results and Result-Files AG70
- Interactive HTML plots (PNG previews): HTML Plot Files shown as PNG
- Age 70 mortality analysis: Age 70 Mortality Analysis Results
- Simple explanation: Peircean Evidence‑Weighted RMST – Simple Explanation
- Method paper / technical details: Peircean Evidence‑Weighted RMST – Method Paper
- IPW RMST Stress-Test Simulation: Clone Censor weight RMST Stress-Test simulation