Bone Cancer Survivability Prognosis with KNN and Genetic Algorithms

Abstract

Cancer survivability or survival optimization models are crucial for treatment programs in hospitals. Predictive models can enable primary care physicians in providing more informed decisions specifically in evaluating proper probability attributes (risk) in relation to outcome (impact) and subsequently the overall result (expected outcome) of treatment procedures. Predictability on survival in health care is very much related to a decision-making process. Better survivability prediction can help doctors decide what type of palliative care is needed and when to commence the treatment. The remaining lifetime of patients can also result in the patient living a fuller and fulfilling life. In this paper, we examine survivability optimization prediction for patients who are diagnosed with osteosarcoma (also known as bone cancer). Survivability prognosis is a statistical predictive approach used to forecast probable recovery. A common statistical approach used for this purpose is known as the Kaplan-Meier analysis. Survival predictability has become integral to the facilitation of patient care and resource optimization. This paper aims to provide an alternative survivability assessment using K-nearest neighbor (KNN) and genetic algorithms.