Disease multiple exposure hierarchical and penalization models

Occupational epidemiological studies investigating health outcome associations with single exposures have contributed to our current knowledge on occupational factors related to various diseases. However, this approach is hypothesis-driven and limits agnostic exploration of unknown disease factors. As a proof of principle, we used hierarchical and penalization models to explore occupational risks associated with lung cancer while accounting for exposures to multiple known carcinogenic exposures in a pooled lung cancer case-control population. Models developed in this work may be readily applied to the EPHOR mega cohort to agnostically explore new exposure-disease associations. Model outcomes may also be used to evaluate and improve exposure estimates by EURO-JEM. Samples codes for hierarchical and penalization models approaches are available on Github (https://github.com/cb-ge/ephor_wp4_agnostic_models) or can be downloaded below.