Abstract:
Osteoporosis is a major global health concern, particularly in developed countries, where it causes bone fragility and increases fracture risk. Affecting over 200 million people worldwide, the disease places a significant economic burden on healthcare systems. Early and accurate diagnosis is essential for effective management; however, existing diagnostic approaches largely rely on Bone Mineral Density (BMD) measurements or X-ray imaging alone. This dependence on single data sources limits diagnostic accuracy and often leads to delayed or missed diagnoses, highlighting the need for more comprehensive solutions.
This research proposes a multichannel osteoporosis prediction system that integrates clinical healthcare data with X-ray images using advanced machine learning techniques. Clinical data are analysed using a Random Forest classifier, while X-ray images are processed through a deep learning model. A voting-based fusion mechanism combines predictions from both channels, enabling a more robust and reliable diagnosis by leveraging complementary information from clinical and imaging data.
Preliminary results indicate that the proposed system outperforms traditional single-method approaches in terms of precision and reliability. The integrated framework offers an interpretable, scalable, and accurate diagnostic solution suitable for clinical use. Future work will focus on optimising data fusion strategies and expanding the dataset to further enhance predictive performance. This study provides a strong foundation for improving osteoporosis detection and management, leading to faster and more accurate diagnostic outcomes.
