Overview
This study builds AI models that score diagnostic screening tests, and that predict screening results, clinical judgment, and life expectancy. Longevity Metrics collects a battery of clinical tests on each participant, in whole or in part, and follows every participant for life.
The sit-to-rise test and the timed walk are scored by hand today, from a person's count. A model scores the same test from video instead. It also measures what no one can count by eye - speed, asymmetry, steadiness - so one capture yields both the original score and additional measurements, intended to enrich the model and strengthen what it predicts.
Every test in a participant's record measures the same body, so the tests are correlated: a test that was performed carries information about one that was not. A model trained across the library learns those relationships and estimates a missing result from the results that are present. Each estimate is checked against records where that part was actually measured, and over decades against death and disease through linkage to the 100-Year Human Aging Study (NCT07563777).
The hypothesis is that the full battery can eventually be predicted across modalities with high accuracy using a few short video clips, replacing most in-person screening. That would let preventive screening reach people and places a physical laboratory cannot. How far the input can be reduced is the question this study exists to answer.
Every model is a physician-reviewed clinical decision aid until it is cleared by the FDA.
Description
The models serve three aims. First, they automatically score simple physical and cognitive tests that already predict function and in some cases mortality, such as the sit-to-rise test and the timed walk. A model reads richer detail from the same recording than a human scorer can, so it improves on the human score rather than only reproducing it.
Second, they predict the parts of a screening a participant did not obtain from the parts that were performed, and increasingly from inexpensive standardized inputs such as a short video. Within a single record, every test is correlated with the others, so each test can both predict the ones that were not performed and serve as the truth against which those predictions are checked. A missing-data engine fills any missing part of a record by leave-one-out across the library.
Third, they predict the physician's clinical judgment where no determining measurement exists.
Models are developed by milestone freezing with forward validation. No model is validated on records it was trained on. Each model is validated in two stages. It is first validated against a human scorer for measurement accuracy, which gates its use as a clinical decision aid. It is then validated over decades for what it predicts about death and disease.
The platform's distinguishing asset is mortality. Every participant is followed for life, so a small library with verified death outcomes answers questions a much larger library without them cannot. The library is the durable asset, studied across geography and time by increasingly capable models.
This study is one of four that compound into one system. The 100-Year Human Aging Study (NCT07563777) supplies the clinical data and validates what it means for health, disease, disability, and death. The Human Observatory Study (NCT07646782) does the same with sociodemographic and environmental data, and receives each model's geographic residuals. The Health Ahead Comparative Effectiveness Study (NCT07669168) moves the screening toward increasing automation and mobility while maintaining quality. This study builds the models that make automation, prediction, and broad utilization possible.
A physician or licensed provider reviews and signs every result a participant receives.
Eligibility
Inclusion Criteria:
- Age \>= 18
- willing to participate in the study
Exclusion Criteria:
- Age \< 18 years
