Hypothesizing precise cancer treatments based on patient survival using machine learning & deep learning

Abstract

Cancer, an enduring medical enigma with historical recognition dating back to ancient civilizations, remains without a definitive cure. This research undertakes a comprehensive investigation encompassing nine prevalent global cancer types, including those with significant implications for the population of Bangladesh. Employing cutting-edge machine learning (ML) and deep learning (DL) models, as well as traditional machine learning techniques, our study derives its strength from an extensive dataset sourced from the Surveillance, Epidemiology, and End Results (SEER) program. Our research endeavors to unravel the intricate tapestry of cancer by distilling pivotal insights from substantial datasets. At its core, our mission is to redefine the landscape of cancer treatment through the creation of predictive models, thus heralding an era of personalized and highly efficacious cancer therapies. Based on a hypothesis, our objective seeks to improve cancer treatment by developing predictive models. Through a comparative analysis involving traditional machine learning models, deep learning algorithms, and boosting models, we have discovered that the boosting models stand out in terms of accuracy, indicating their potential to enhance predictive precision for therapeutic response. We hypothesize that the surgical removal of localized tumors can effectively arrest cancer progression, thereby increasing patient survival. This encapsulates the main focus of our study, which is a committed attempt to identify unique answers to a persistent medical dilemma by integrating the knowledge of the past with the potential of the future.