A new approach to prediction of radiotherapy of bladder cancer cells in small dataset analysis

Bladder cancer is a common urologic cancer. Radiotherapy plays an increasingly important role in treatment bladder cancer due to radiotherapy preserves normal bladder function. However, the five-year survival rate after radiotherapy for bladder cancer patients is 30–50%. Some biological proteins influence the outcome of radiotherapy. One or two specific proteins may not be sufficient to predict the effect of radiotherapy, analyzing multiple oncoproteins and tumor suppressor proteins may help the prediction. At present, no effective technique has been used to predict the outcome of radiotherapy by multiple protein expression file from a very limited number of patients. The bootstrap technique provides a new approach to improve the accuracy of prediction the outcome of radiotherapy in small dataset analysis. In this study, 13 proteins in each cell line from individual patient were measured and then cell viability was determined after cells irradiated with 5, 10, 20, or 30 Gy of cobalt-60. The modeling results showed that when the number of training data increased, the learning accuracy of the prediction the outcome of radiotherapy was enhanced stably, from 55% to 85%. Using this technique to analyze the outcome of radiotherapy related to protein expression profile of individual cell line provides an example to help patients choosing radiotherapy for treatment.

Research highlights
► A useful model of bootstrap technique to analyze the outcome of radiotherapy related to protein expression profile of individual cell line.
► Increasing the number of training data enhances the learning accuracy of the prediction the outcome of radiotherapy.
► A new approach is used to improve the accuracy of prediction the outcome of radiotherapy in small dataset analysis.