Mammography is a valuable tool for reducing breast cancer mortality, however the frequency with which women should receive mammography is still controversial, and guidelines for screening mammography vary across different jurisdictions. Micro-simulation modeling has the potential to aid health policy researchers in evaluating the impact of a policy change (and other changes) at both the individual level and the population level. Micro-simulation modeling has two components: a natural history model and an intervention model.


To create a micro-simulation model, the natural history of the disease must first be modelled. Several previous studies have created mathematical models of breast cancer growth; however, these models required assumptions be made about certain aspects of cancer development, e.g. the distribution of tumour growth rates2-5 . In order to minimize the need for assumptions, this project will use a technology that has yet to be utilized in breast cancer modelling: computer-aided detection (CAD) software. CAD is a system that aids radiologic interpretation of mammograms by highlighting areas on breast images that display suspicious lesions, and assigning a level-of-certainty score to each of those regions.


We propose that by comparing abnormal CAD scores (taken from the abnormal mammogram that led to breast cancer detection) with normal CAD scores from previous normal mammograms of women who have developed breast cancer, we expect to be able to create a preliminary natural history model that is a more realistic representation of the biological development of screen-detected breast cancer than existing models. This natural history model would inform the microsimulation model currently under development at the BC Cancer Agency, which will ultimately be used to evaluate preferable screening frequencies for women in British Columbia and elsewhere.