MPIDR Working Paper

Predicting individual-level longevity with statistical and machine learning methods

Badolato, L., Decter-Frain, A. G., Irons, N., Miranda, M. L., Walk, E., Zhalieva, E., Alexander, M., Basellini, U., Zagheni, E.
MPIDR Working Paper WP-2023-008, 26 pages.
Rostock, Max-Planck-Institut für demografische Forschung (Februar 2023)
Open Access


Individual-level mortality prediction is a fundamental challenge with implications for people and societies. Accurate longevity predictions improve life planning, targeting of high-risk individuals, and organization of social interventions, policies, and public spending. Demographers and actuaries have been primarily concerned with mortality modeling and prediction at a macro level, leveraging strong regularities in mortality rates over age, sex, space, and time. Besides clinical settings, individual-level mortality predictions have been largely overlooked and have remained a challenging task. We model and predict individual-level lifespan using data from the US Health and Retirement Study, a nationally representative longitudinal survey of people over 50 years of age. We estimate 12 statistical and machine learning survival analysis models using over 150 predictors measuring behavioral, biological, demographic, health, and social indicators. Extending previous research on inequalities in mortality and morbidity, we investigate inequalities in individual mortality prediction by gender, race and ethnicity, and education. Machine learning and traditional models report comparable accuracy and relatively high discriminative performance, particularly when including time-varying information (best mean Area Under the Curve = 0.87). However, the models and predictors used fail to account for a majority of lifespan heterogeneity at the individual level. We observe consistent inequalities in mortality predictability and risk discrimination, with lower prediction accuracy for men, non-Hispanic Blacks, and low-educated individuals. In addition, people in these groups show lower accuracy in their subjective predictions of their own lifespan. Finally, we see minimal variation in the top features across groups, with variables related to habits, health history, and finances being relevant predictors. Our results assess how well mortality can be predicted from representative surveys, providing baselines and guidance for future research across countries.

Schlagwörter: Vereinigte Staaten, forecasts, inequality, longevity
Das Max-Planck-Institut für demografische Forschung (MPIDR) in Rostock ist eines der international führenden Zentren für Bevölkerungswissenschaft. Es gehört zur Max-Planck-Gesellschaft, einer der weltweit renommiertesten Forschungsgemeinschaften.