Longevityresearch.ca saves up to 8.6 life years/patient with NHANES 1988–2018 series — 101,316 persons

TORONTO - June 16, 2026 - PRLog -- Where the modeled headroom concentrates

The modeled gain is largest where current standard of care leaves the most modifiable post-acute risk on the table. Pulmonary disease (COPD/IPF/PAH) showed the largest mean gain at roughly +11.0 life-years, followed by heart disease (cardiovascular) at about +10.2 — the two areas the initiative flags as having the most room for improvement over standard care. Liver disease (+9.5) and metabolic disease/type-2 diabetes (+9.5) followed, with brain (+7.1) and cancer (+5.7) smallest; cancer mortality sits largely in an acute or age-driven phase that no prevention bundle can reach.

A separate, deliberately conservative comparison underscores the point. When the harness measures the optimum against the drugs patients were actually prescribed — the overlap-free view — the mean prescribed risk reduction is only about 7%, against a modeled Pareto ceiling near 73%, leaving a roughly 66% unrealized gap. Much of that gap reflects lifestyle and procedural interventions that never appear in a prescription record at all.

A working tool, available now
The harness is live and runs entirely in the browser at longevityresearch.ca/mimic_mortality_harness, with no data upload and no server round-trip — files are parsed locally, so the analysis is private by construction. It is open for anyone to run their own data. It accepts MIMIC-III clinical tables (CSV) and NHANES data in several forms — SAS-transport (.XPT) files, the harmonized 1988–2018 CSV modules used in this test (available openly via figshare; Nguyen et al., NHANES 1988–2018: https://figshare.com/articles/dataset/NHANES_1988-2018/21...), or NHANES III fixed-width files (.dat paired with their .sas layout) — and the initiative will consider adding further input formats on request. The pipeline streams large files column-by-column, so survey datasets with a thousand-plus columns load without overwhelming the browser, and it returns the same standard-of-care-versus-Pareto results at full cohort scale.
About the Longevity Research Initiative

The Longevity Research Initiative is an independent quantitative-research project applying structural causal models (Judea Pearl's framework), backdoor adjustment, cross-correlation removal, E-values, and dose-response saturation to the question of how much of premature mortality is modifiable. Its open "Bayesian Causal Atlas" hosts more than 30 interactive clinical analyses, each paired with a formal report. All outputs are framed as upper-bound ceilings for validation rather than clinical guidance.

Contact: Longevity Research Initiative · longevityresearch.ca · [stuart@longevityresearch.ca]

Resources: Live harness — longevityresearch.ca/mimic_mortality_harness · Dataset — figshare.com/articles/dataset/NHANES_1988-2018/21743372