Author: Sierra Wagner Major: Industrial Engineering 

During the Fall 2019 and Spring 2020 semesters, I performed research in the industrial engineering department under my advisor, Dr. Shengfan Zhang. I was thrilled to have the opportunity to work with Dr. Zhang because of her extensive experience in breast cancer research and expertise in simulation and modeling for medical decision making. Dr. Zhang introduced me to the topic of breast cancer over-diagnosis, which is considered as one of the significant harms of breast cancer screening.  

Breast cancer over-diagnosis risk is difficult to estimate and varies significantly across current research. Therefore, the overall goal of my research was to understand the impact that the range of breast cancer over-diagnosis rate estimates in the current literature have on patient outcomes. My specific goal was to establish a simulation approach to examine the relationship between breast cancer over-diagnosis, breast cancer treatment policies, and patient outcome.  

I spent the first half of the Fall 2019 semester doing background research on breast cancer disease progression and over-diagnosis and reviewing current literature which was relevant to modeling disease progression, screening, treatment, and over-diagnosis. Once I had a grasp on the field of study, Dr. Zhang and I spent the rest of the Fall 2019 semester formulating the framework which I would use in my simulation model. 

The simulation framework that we developed integrates two components which we identified as key for modeling disease progression and the cancer detection process. The first component, representing disease progression, is based upon a Markov chain model (Ross, 2014) that was devised to describe patient health state and changes over time. The second component, representing the cancer detection process, models screening practices or policies. The patient health state, determined by the first component of the model, acts as a signal to the screening component of the model. 

At the start of the Spring 2020 grant period, I was ready to begin building my simulation model. Constructing the simulation model was a relatively smooth process since I had a solid understanding of the framework that Dr. Zhang and I had formulated. However, I struggled to understand or come up with an idea for how we would incorporate over-diagnosis into the simulation model. Dr. Zhang proposed that we represent over-diagnosis in the model by incorporating a set of disease regression probabilities, and the final pieces of the simulation fell into place.  

Unfortunately, as I was finishing up my simulation model and as it was time to perform numerical experiments and sensitivity analysis to draw conclusions from our research, the semester was interrupted by COVID-19 and Dr. Zhang and I were no longer able to meet in person. Despite this challenge, Dr. Zhang remained helpful as ever and provided me with the guidance I needed to conduct the numerical experiments and sensitivity analysis necessary to draw meaningful conclusions about the impact of our parameter inputs on patient outcome. We found that treatment policy was statistically significant with respect to patient outcome and that over-diagnosis risk was partially statistically significant in its impact on patient outcome. 

Through this Honors College Research Grant, I was able to learn about breast cancer, the harms of over-diagnosis, and modeling techniques for medical decision making. I was able to independently apply simulation techniques which I learned in the classroom to a complex process, and I am very proud to have successfully performed statistical analysis which will contribute to breast cancer research.