eCPC WP5: Prediction and screening models

This project will develop predictive and screening models for breast and prostate cancer on several levels.

Breast and prostate cancer will be modeled on several levels including probabilistic modeling of tumor initiation, characterization and growth and how these depend on a large and complex system of demographic, molecular and background variables. The underlying processes are multidimensional and time dynamic and only partially observed. Detection, prevention and treatment strategies will be imposed on the system and evaluated using existing data (WP1, WP2) and the in silico microsimulation engine (WP3). Tools to be developed and used include probabilistic prediction models, learning methods, hierarchical Bayesian networks and and tools in systems biology (WP4). Note in particular that to evaluate screening strategies it is important to jointly model growth rate and screening test sensitivity. Due to data incompleteness a smooth analytic tumor growth rate curve is needed that accounts for larger growth for small tumors (governed by cell reproduction rate of a given tumor cell) with growth velocity gradually leveling off with tumor size due to e.g. more limited nutrition for the tumor.

Project link

Investigators

PI: Keith Humphreys
Co-PI:
Member: Martin Eklund
Member: Alexander Ploner
Member: Mark Clements