Osteoporosis Disease Prediction Using Machine Learning Algorithms

Osteoporosis is a long-term bone condition in which reduced mineral density weakens skeletal strength and increases susceptibility to fracture-related complications. In addition, accurate and early identification of osteoporosis can significantly decrease disease-related complications and enhance overall patient outcomes. This paper compares three machine learning methods without tree-based approaches, namely Logistic Regression, Support Vector Machine, and K-Nearest Neighbors, for the prediction of osteoporosis disease. Feature normalization and standard scaling were performed as preprocessing techniques to optimize model performance. Accuracy, precision, recall, F1-score, and receiver operating characteristic-area under the curve metrics are considered to evaluate the efficiency of the proposed models. Experimental results illustrate that the Logistic Regression model achieved the highest prediction accuracy, 98.56%, and outperformed other models in disease prediction. Support Vector Machine showed very stable and robust classification performance with an accuracy of 98.12%. On the other hand, K-Nearest Neighbors provided moderate performance with an accuracy of 94.13%. Thus, it is clear that machine learning models are both effective and computationally efficient in the prediction of osteoporosis disease and support clinical decision-making systems.