Machine Learning Methods for Prediction of Brain Tumors and Pneumonia Diseases

Abstract

Pneumonia and brain tumors are considered critical diseases due to the substantial challenges related to accurate prediction and diagnosis at an early stage. Machine learning (ML) methods are used in medical imaging processing to detect specific patterns and features within input images and automatically classify various medical conditions. This paper aims to predict and classify pneumonia and brain tumors diseases, to compare the ML performance of methods: Decision Tree (DT), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Random Forests (RF), Logistic Regression (LR), and Naïve Bayes (NB), and to analyze the impact of dataset increasing size on the classification performance. This study reveals that the Random Forest algorithm achieves the best performance, with 90% accuracy in the brain tumors dataset and 79% accuracy in pneumonia disease prediction.