WP3: Cancer microsimulation

New screening strategies driven by genetic markers and molecular profiles on top of traditional mammography and PSA measures are expected to increase cost-effectiveness by improving screening specificity (reduce rate of false positives) without jeopardizing screening sensitivity (ability to detect tumors). The system against which to evaluate screening strategies is expected to be too complex to be captured by a single analytic probabilistic model. Instead this project develops a continuous time stochastic simulation engine for interconnecting prediction models for cancer occurrence and progression with strategies for early detection, prevention and treatment. A generic system will be set up with separate implementations for breast and prostate cancer. The model is denoted the CRisP Screening Simulation, CRisPSS and is driven by observed data from the CRisP cohorts KARMA and STHLM2 and by external knowledge on cancer risk and progression. CRisPSS builds on a set of probabilistic sub-models for the risk factor distribution and for tumor characteristics, growth and detection. CRisPSS will be designed for model update as information from projects 4 and 5 and from external sources accumulate. Model updates are evaluated against detection rates and distribution of tumor size and characteristics at detection.