Enhancing Healthcare Analytics: A Multiclass Supervised Learning Approach Approach for Disease Classification

Due to modern climate change and lifestyle choices, humans are increasingly vulnerable to a wide range of diseases. The early detection and forecasting of such diseases are of great significance in order to mitigate their severity. The precise identification of diseases by doctors by manual means is often challenging. This paper aims to ascertain and forecast the individuals suffering from prevalent long-term illnesses based on the symptoms. The reliable identification of individuals with diseases can be accomplished by employing a machine learning approach. This paper presents a comprehensive analysis of the novel proposed method, which incorporates pre-processing of the dataset and training the classifiers. The dataset utilized in this study is a widely accessible Kaggle dataset, which encompasses 41 distinct classes and 133 symptoms. In this study, five distinct machine learning (ML) techniques are employed in conjunction with a 10-fold cross-validation approach to discern and classify a total of 41 diverse disorders effectively. 100% testing accuracy is achieved by the three models: Support Vector Machines, Random Forest, and Decision Trees. The recommended approach serves as a critical component in facilitating healthcare workers in making effective and well-informed decisions, consequently improving disease recognition, diagnosis, and therapy planning within the medical field.