Reliability Models in Cancer Epidemiology

A special class of reliability models called “cumulative damage models” is introduced and demonstrated that it very well matches the view of epidemiology on carcinogenesis. This new approach helps to overcome conceptual problems of existing models and, moreover, allows among other things the incorporation of mathematical representatives for host responses which has not been achieved so far by other models. The paper outlines first results concerning descriptive epidemiolo‐gical data as well as data on special exposure histories or applications to dose‐response considerations in animal experiments. The new mathematical issue implies partially substantially differing interpretations of observed data, compared to existing models, and promises that cumulative damage models will be a powerful tool for theoretical epidemiological investigations in the future.