Generalizability to civilian populations

Determine whether the stochastic, stress-modulated compartmental stochastic differential equation model parameterized using U.S. veteran electronic health record covariates generalizes to civilian populations with different psychosocial risk patterns.

Background

The model is constructed and parameterized using covariates and effect sizes derived from U.S. veteran health records. Veterans have distinct psychosocial risk profiles compared to civilian populations, potentially limiting external validity.

Assessing generalizability is necessary to understand whether the model’s dynamics and risk predictions remain reliable outside the veteran population and to support broader public health applications.