| Osteoporosis is characterized by low bone mass and deterioration of bone tissue caused by altered bone microstructure which leads to increased bone fragility and further results in susceptibility to fractures. It is the most frequent metabolic bone disease to date that poses a major global public health problem due to its high morbidity, caused by fractures in the older population. Osteoporosis typically becomes symptomatic after a fracture has occurred, which leads to an increasing financial burden on the health care system alongside secondary complications. Early detection of increased risk for osteoporosis is crucial in reducing the likelihood of these complications. In this study, several machine learning models were examined to predict osteoporosis based on clinical data and laboratory results. Among the machine learning models used, Logistic Regression produced the best predictive capability, indicating that the model is reliable in predicting osteoporosis cases. The most important attributes contributing to this risk based on explainable artificial intelligence were features such as weight, total cholesterol, age, relative lymphocytes, low-density lipoprotein cholesterol, alkaline phosphatase, and direct bilirubin. The adequacy of predictive capacity clearly demonstrates the easibility of integrating machine learning models into clinical practice to provide valuable diagnostic and prognostic support to both physicians and patients. This contributes to an enhanced decision-making process in identifying patients at risk and thus early intervention and personalized treatment plans can be created in the process. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.