Blockchain and Deep Learning for Smart Healthcare
by Akansha Singh, Anuradha Dhull, Krishna Kant Singh
16An Improved Random Forest Feature Selection Method for Predicting the Patient’s Characteristics
K. Indhumathi1* and K. Sathesh Kumar2
1Department of Computer Application, Kalasalingam Academy of Research and Education, Kalasalingam University, Krishnan Kovil, India
2Department of Computer Science and Information Technology, Kalasalingam University, Krishnan Kovil, India
Abstract
The global population has been devastated by the Coronavirus Disease 2019 (COVID-19) epidemic that started in Wuhan, China, which has overwhelmed established medical systems globally. In March of this year, there had been over 116 million infected patients along with 2.58 million deaths worldwide. Day by day, the number of confirmed COVID-19 patients grows. However, to provide efficient treatment, expert system approaches must be applied to predict the outcome of an infected patient and the increase in the number of patients. Preprocessing the patient health record and extracting the essential features from the preprocessed health record are the first steps to determine the probability of infected people. This chapter proposes a new modified algorithm for predicting the best features of a patient health dataset. It also discusses various feature selection methods and analyzes the performance of different feature extraction methods. To extract important features from health records, this chapter uses the Boruta algorithm, Rank features by importance, Recursive feature elimination, Variable importance ...