A Binary Survivability Prediction Classification Model towards Understanding of Osteosarcoma Prognosis

Saravanan Muthaiyah (Faculty of Management,Multimedia University, Selangor,)
Vivek Ajit Singh (Faculty of Medicine, Universiti Malaya, Selangor,)
Thein Oak Kyaw Zaw (Faculty of Management,Multimedia University, Selangor,)
Kalaiarasi S. M. Anbananthen (Faculty of Information Science and Technology, Multimedia University, Melaka,)
Byeonghwa Park (Department of Management and Marketing, Valdosta State University, Georgia,)
Myung Joon Kim (Department of Big Data Application, Hannam University, Daejeon,)

Emerging Science Journal
Vol 7, No 4 (2023): August

The objective of this study is to explore effective and innovative machine learning techniques that can assist medical professionals in developing more accurate prognoses that can enhance the survivability of osteosarcoma patients by investigating potential prognostic factors and identifying novel therapeutic approaches. A comprehensive analysis was conducted using a dataset of 128 osteosarcoma patients between 1997 to 2011. The dataset included 52 attributes in total that covered a wide range of demographics, together with information on clinical records, treatment protocols, and survival outcomes. Data was obtained from NOCERAL (National Orthopaedic Centre of Excellence in Research and Learning), Kuala Lumpur. Three distinct binary classification methods (i.e., random forest, support vector machine (SVM), and artificial neural network (ANN)) were employed to identify the prognostic factors that are associated with improved survival efficacy measures. The results of this study revealed that both SVM and ANN outperformed random forests in predicting survivability for both the 2-year and 5-year time frames. These findings indicate the potential of SVM and ANN as effective tools for predicting osteosarcoma survivability. The study signifies a significant step towards integrating machine learning techniques into the existing toolkit available to medical practitioners. This study contributes to the medical field by providing a comparative analysis of three prominent machine learning techniques for predicting osteosarcoma survivability. The superior performance of SVM and ANN over random forests highlights the potential of these methods in generating more accurate survivability predictions. Further development and refinement of these machine learning techniques hold promise for enhancing their effectiveness and instilling greater confidence among medical professionals and patients in the predictive capabilities of machine learning and artificial intelligence models for osteosarcoma survivability. Doi: 10.28991/ESJ-2023-07-04-018 Full Text: PDF