Examination of Treatment Outcome on Brain Tumour Using Large Sample Data of Wisconsin Diagnostic Centre

Abstract

This study uses a multinomial logistic regression model to examine the treatment outcome of brain tumour parameters. Wisconsin data centre is the source of the study's sample (diagnostic data). Nine variables and 2000 observations make up the data collected from Wisconsin data centre. Python programming software was employed for the analysis. The boxplot was employed to ascertain the frequency of each variable in terms of first, third and median quantile values.  The correlation coefficient result was used to measure the relationship between the nine variables. The relationship between tumour grade and tumour location shows weak correlation, with values of 0.333, whereas the correlation between time to recurrence (month) and survival time is 0.42. The strong correlation between tumour grade and recurrence site for treatment is 0.99. However, other variables do not correlate effectively. Given the stages of treatment outcome, we use the treatment outcome as the independent variable to recast the relationship between the variables. The probability value was used to calculate the impact of the independent variable on the dependent variable. Treatment outcome, given a partial response on tumour grade, has no bearing on survival time (month). The other variables are impacted by the partial response to the treatment. The progressive stage treatment outcome demonstrates that while gender and survival time are unaffected by a complete response, all other factors are affected. Then we observed that on the complete response nature of the disease, there is no significant difference in treatment results across all brain tumour parameters. Treatment outcome given partial treatment has a considerable impact on the remaining variable, however, there is no significant difference between treatment outcome of gender on the effect of tumour type and survival time.