Enhanced Prognosis Prediction Using Ensemble Machine Learning: A Comparative Study

Abstract

Clinical predictive models have become more significant in modern treatment because they educate doctors, healthcare workers, patients, and their families about potential outcomes, hence facilitating clinical and improving the condition of patients. Diagnostic forecasting methods try to determine a person’s probability of currently suffering from an illness, whereas prognostic forecasting methods seek to estimate the risk-related probability of developing specific illnesses developing in the years to come. Prognosis prediction is essential for improving clinical decision-making and patient outcomes. This work presents a comparative analysis of five supervised learning algorithms- viz., random forest, decision tree, naïve bayes, Support Vector Machine (SVM), and K-Nearest Neighbors (KNN) for predicting disease outcomes (prognosis) using clinical dataset. The generated clinical dataset contains a diverse collection of 113 unique diseases across several levels of severities, detailing diseases-related information/ relationships, symptoms, patient demographics, and medical indicators. The exploratory data investigations revealed many such correlations/ relationships between multiple factors contributing to a disease. Experimental results, based on random test evaluations, show that the Random Forest model achieves a superior accuracy of $96.2 \%$ along with high precision, recall, and F1-score. Furthermore, recent IEEE studies report accuracies below $96 \%$, indicating that our approach offers a significant improvement.

Ähnliche Arbeiten