All Times EDT
Keywords: Data driven; Machine learning techniques; Statistical modeling; Multiple myeloma cancer; Health science
Multiple Myeloma (MM) cancer has been and continues to be the subject of many research studies. The main goal is to improve the therapeutic/treatment process of the survival of MM patients. Based on the 2012-2016 MM cases and deaths, the number of new cases was 6.9 per 100,000 men and women per year, and the number of deaths was 3.3 per 100,000 men and women per year. It is therefore imperative to research MM. In the present study, we proposed a data-driven statistical model for the survival time of 48 patients diagnosed with multiple myeloma as a function of 16 attributable risk factors. We are interested to know the causes of MM and the life span of MM patients given the risk factors associated with it. We identified 9 attributable risk factors out of 16 and one interaction term to be significantly contributing to the survival time. They are Bence Jone protein in urine, blood urea nitrogen (BUN)/serum creatinine, infections, % myeloid cells in peripheral blood, fractures, serum calcium, gender, platelets and age, and white blood cells & total serum protein an interaction term. The proposed model satisfied all the model assumptions, passes the residual analysis test and has very high prediction accuracy. Thus, it passes the goodness-of-fit test and the qualities of a good model. The identified significant attributable risk factors and the interaction has been ranked based on the percent contribution to the survival time. The proposed model was evaluated and compared with other existing models of survival of multiple myeloma. Our model is very accurate and also identifies some new significant risk factors. The study offers an improved strategy for the therapeutic/treatment process of multiple myeloma cancer.