DMS Graduate Student Seminar

Speaker: Serhat Simsek, a PhD student in our department, specializing in statistics; his advisor is Dr. Mark Carpenter.

Title: Do Analysts Mislead Medical Practitioners? A Comprehensive Analytics Technique to Better Detect Non-Surviving Cancer Patients

Abstract: Analysis of survival times of cancer patients is crucial for medical practitioners to determine possible outcomes and make better future-plans for the patients. In the healthcare analytics literature, it is common to see employment of machine learning algorithms to predict survivability of the cancer patients. In this study, we detected the common misleading methodology that has been used in the literature in predicting the surviving and non-surviving cancer patients and propose a comprehensive modeling technique that overcomes the issue of the misprediction of survival. In order to illustrate the issue and its solutions, we deploy Artificial Neural Networks, Random Forest and Logistic Regression. The comprehensive model is applied to rectum cancer data and results are validated with breast cancer data from the SEER. The results will assist medical practitioners to make better decisions for their patients and thus appropriate interventions.