Cancer Care Modelling with Micro-Simulation

–Ongoing Projects–


Breast Cancer Care Pathway and Micro-simulation Model

In recent years, modeling cancer in populations has led to improved screening and prevention initiatives as there are many issues in breast cancer early detection and treatment that lack conclusive support from published studies. Currently it is not possible to evaluate the full impact of these screening or treatment interventions for the province of British Columbia. Therefore, we are developing a breast cancer micro-simulation model that will be used to further improve breast cancer screening and treatment interventions and strategies.

A comprehensive map of the breast cancer care paths used in British Columbia was developed, including early detection, diagnosis, treatment (surgery, chemotherapy, and radiotherapy), survivorship and end-of-life care. This map forms a framework from which a micro-simulation model of breast cancer in BC will be developed and validated.  This framework was then populated and validated with screening and treatment data for individuals diagnosed between years 2001-2005. This step of populating our platform with data is a gold standard in population modeling, and will allow a comprehensive epidemiological model to be developed and validated for the BC population. Future work will also involve integration of screening and treatment costing data into the model.

As a final product, the model will evaluate the impact of screening and other treatment interventions on downstream factors such as population health and costs to the healthcare system. Availability of such a breast cancer model will enable researchers and policy makers to evaluate the effects and effectiveness of different interventions before they get implemented, with particular emphasis on individual risk-based personalized screening methods.


Breast Cancer Outcomes Dashboard

Cancer patients can present with an uncommon set of disease characteristics, and traditional bodies of evidence often provide limited insight for estimating outcomes. However, with the increased collection of demographic, disease, and outcomes data, it is possible to build simplistic data visualization tools which give clinicians and researchers the ability to examine trends in oncologic outcomes for specific sub-populations of interest. 

The Breast Cancer Outcomes Dashboard is a proof of concept software designed to demonstrate the utility of giving end-users (i.e. oncologists) the ability to select cohorts of breast cancer patients and view their survival outcomes. In other words, it is an offshoot project meant to create a tool that allows non-technical users (such as physicians) the ability to select cohorts of patients from within our 5-year data set and have survival statistics (among other data) calculated for them on the fly. The current dashboard includes parameters such as: Age of diagnosis, Diagnosis Date, Cancer Site, TNM Staging, Histology, etc. Certain parameters can have multiple items selected (i.e. various T stages). Each parameter can be selected independently, and for up to two cohorts. Queries are then built for each cohort, and a Kaplan-Meier survival curve is created for both populations. The curves include censoring of patients (based on 10 year survival). A log rank chi square test is performed to compare the significance of the two curves, and is presented as a P-value. 

In future, we plan to expand this tool to include datasets for other cancers, with new data being automatically imported every 6 months.