Feature selection (FS), a crucial preprocessing step in machine learning, greatly reduces the dimension of data and improves model performance. This paper focuses on selecting features for medical data classification. Methods: In this work, a new form of ensemble FS method called weighted rank difference ensemble (WRD-Ensemble) has been put forth. It combines three FS methods to produce a stable and diverse subset of features. The three base FS approaches are Pearson’s correlation coefficient (PCC), reliefF, and gain ratio (GR). These three FS approaches produce three distinct lists of features, and then they order each feature by importance or weight. The final subset of features in this study is chosen using the average weight of each feature and the rank difference of a feature across three ranked lists. Using the average weight and rank difference of each feature, unstable and less significant features are eliminated from the feature space. The WRD-Ensemble method is applied to three medical datasets: chronic kidney disease (CKD), lung cancer, and heart disease. These data samples are classified using logistic regression (LR). Results: The experimental results show that compared to the base FS methods and other ensemble FS methods, the proposed WRD-Ensemble method leads to obtaining the highest accuracy value of 98.97% for CKD, 93.24% for lung cancer, and 83.84% for heart disease. Conclusion: The results indicate that the proposed WRD-Ensemble method can potentially improve the accuracy of disease diagnosis models, contributing to advances in clinical decision-making.; February 2024

Integration of Internet of Healthcare Things (IoHT) and Machine Learning (ML) can be used successfully in healthcare systems to increase the accuracy of computer-aided diagnosis. This paper emphases on the application of IoHT and ML in detecting spinal abnormalities, which can be integrated with IoHT. The novelty of this work is in the use of multilayer perceptron (MLP) to a spinal dataset to obtain high accuracy in spinal abnormality detection. The dataset has 310 samples and is freely available on Kaggle repository. We use Pearson correlation coefficient (PCC), ReliefF and Gain ratio (GR) filter-based feature selection methods to select the top 10, 8, 6 and 5 features according to relevance or weight of features in preprocessing stage. In classification stage, we use logistic regression (LR), support vector machine (SVM), and Bagging algorithm in addition to MLP. The experimental results indicate that the PCC feature selection technique and MLP classification algorithms give the most promising results. A maximum classification accuracy of 88.0645% is obtained when MLP is used after selecting the top 5 spinal features by PCC feature selection method. This accuracy obtained by MLP and PCC is higher than 86.96% reported in the literature of spinal disease.; February 2022

; 2023-12-17