MID-LIFE PREDICTORS OF EXTREME LONGEVITY — INTERACTIVE SHAP EXPLORER

Welcome!
This Space lets you explore the models behind our study “Machine learning in epidemiology: an introduction, comparison with traditional methods, and a case study of predicting extreme longevity” (Atias et al. – Annals of Epidemiology, in-press 2025).
DOI 10.1016/j.annepidem.2025.07.024

What you can do

• Visualise how each mid-life risk factor pushes longevity probability up or down.
• Compare three models: logistic regression, LASSO and XGBoost.
• Inspect individual explanations (waterfall plots) or global patterns (summary plots).

Study snapshot

• Cohort: 10 058 Israeli men aged 40–49 in 1963 (Medalie et al., 1973)
• Follow-up: 5-year CHD surveillance plus full vital-status linkage to 2023
• Outcome: Reaching age ≥ 95 (“near-centenarian”)
• Models: Logistic regression · LASSO (glmnet) · XGBoost
• Explainability: SHAP values for both tree-based and linear models

Data note

Raw individual-level data cannot be shared publicly for privacy reasons.
The app runs on a de-identified snapshot stored securely inside this Space.

Enjoy exploring!