This item is available under a Creative Commons License for non-commercial use only
The idea of long-term survival amongst older individuals has been a major medical and social concern. A wide range of biomarkers have been identified to prospectively predict disability, morbidity, and mortality outcomes in older adult populations. The machine learning techniques applied with clinically relevant biomarkers provide new ways of understanding diseases and solutions to tackle challenges to the health of the aging population. This paper describes two supervised machine learning techniques, Logistic Regression (LR) and Support Vector Machine (SVM) which are used in the prediction of the mortality in elderly people. LR is one of the traditionally used predictive modeling methods in clinical research where the probability of occurrence of two classes is a dichotomous criterion whereas, SVM is an emerging classification supervised learning technique based on building models using maximum-margin hyperplane. An attempt has been made to measure the classifier accuracy of each model and the performance of both the models is compared on a set of biomarker features of old patients. The experimental result shows that the SVM model outperformed the LR model in the prediction of survivorship among old individuals with statistically significant results (p<0.01).
Sonkar, P. (2017) Application of supervised machine learning to predict the mortality risk in elderly using biomarkers. Masters dissertation, DIT, 2017.