Comparative Analysis of Ensemble Techniques for Bone Fracture Detection from X-Ray Images Using HOG Features

Musculoskeletal conditions encompass 150 varying conditions and approximately affect 1.71 billion people worldwide, according to the World Health Organization. Bone fractures are among the musculoskeletal conditions that significantly impact individuals, families, societies, and healthcare systems, as they can lead to work absences, decreased productivity, disability, and a host of other life-limiting conditions. This study aims to evaluate the performance of superior-performing machine learning (ML) models that are fed with histograms of oriented gradients (HOG). The bestperforming ML models were chosen and implemented toward ensemble models in aims to reduce overfitting and address resource-limited settings. The results indicate that by using KNearest Neighbors (KNN), Decision Tree (DT), and Random Forest (RF), which yield the highest scores, the proposed ensemble models were able to increase the score, with the stacking ensemble model achieving accuracy score of 99.30%. Future research could focus on integrating advanced machine learning techniques, hybrid feature extraction methods, and larger datasets to enhance diagnostic accuracy and applicability in clinical environments.