A unified framework links infant vulnerability with aging-related mortality dynamics

A central question in Geroscience is whether early-life mortality, which declines from birth to sexual maturity, and late-life mortality, which grows exponentially in time, can be understood within a shared conceptual framework. We show that stochastic threshold models can explain both phases by incorporating heterogeneity in neonatal vulnerability. Using U.S. National Center for Health Statistics data, we find that infant mortality risk is strongly associated with neonatal clinical markers such as Apgar scores, gestational age, and birth weight, suggesting that initial physiological differences persist across early life. We show that the [~]1/t mortality decline generically arises in stochastic threshold models via depletion of the most vulnerable, across a wide range of model specifications. Incorporating this mechanism into the Saturating-Removal model captures both the early decline and the later Gompertz acceleration, reproducing the full J-shaped mortality curve. Together, our findings link neonatal vulnerability to late-life mortality dynamics within a shared stochastic framework, supporting a life-course perspective on aging and longevity.