Skip to main content
eScholarship
Open Access Publications from the University of California

Analysis of a Bladder Cancer Experiment

  • Author(s): Yin, Kevin
  • Advisor(s): Xu, Hongquan
  • et al.
Abstract

The effectiveness of cancer treatments has always been an issue to both doctors and patients, due to various types of cancer cells as well as their ability to mutate and replicate quickly. In this study, we mainly focus on the analysis of a bladder cancer experiment on the effectiveness of different bladder cancer cell treatments. The experiment has 125 combinations with 8 drugs in total, each with 6-10 dosage levels in result of treating 6 different bladder cancer cells, and the experimenter recorded the percentage of cancer cells remaining after the treatment. The ex- periment was performed in 6 batches. The purpose was to find the optimal drug combination in treating different cancer cells. The lower the percentage remains in a cancer cell, the better the combination is. The analysis procedure involves a variety of models and transformations, including linear model, generalized lin- ear model with binomial, neural network and transformations such as logarithm, square root on the responses, and the results were compared in terms of the R-square, MSE and the actual vs. predicted plots. The best model in terms of the fitness and the prediction accuracy is a second degree linear model with each batch being treated as a block.

Main Content
Current View