Machine Learning System for Asthma Severity Detection Based on Support Vector Machine

Abstract

Detecting the severity of contagious respiratory illnesses early is essential for effective treatment and healthcare planning. This study explores how machine learning, specifically Support Vector Machine (SVM) and Random Forest (RF), can help classify and predict illness severity based on patient symptoms and clinical data. The system is developed using Python programming language on streamlit environment. The aim of this research is to classify the severity level of Asthma for informed decision and patient care. Random Forest regression and SVM models were successfully developed and trained, and its performance was evaluated using metrics such as Mean Squared Error (MSE) and R². The findings showed an MSE of 0.0173 and an R² of 0.6630. These results indicated that the model could predict the severity index of Asthma in a patient with a good degree of accuracy, capturing approximately 88.3% of the variance in the target variable. Our findings show that while both algorithms perform well, Random Forest provides more accurate predictions due to its ability to capture complex patterns in medical data.