622 - Machine Learning Approach to prediction of Osteoporosis based on Clinical Features in Postmenopausal women

Objectives:
Osteoporosis is a systemic skeletal disease characterized by a decrease in bone mass, leading to an increased risk of fractures. The aim of this study was to develop a machine learning-based model to predict the risk of osteoporosis in postmenopausal women based on clinical and circulating biomarker data.

Design:
This study included 447 postmenopausal women from all health centers in Tabriz, Iran. Blood samples were collected from the participants, and eight different machine learning models, including Support Vector Machine (SVM), Random Forest Classifier (RFC), AdaBoost, Gradient Boosting, K-Nearest Neighbor (KNN), Logistic Regression, Decision Tree, and Multi-Layer Perceptron (MLP), were trained using a combination of clinical features and circulating biomarkers. The models were evaluated using performance metrics such as the area under the curve (AUC), accuracy, recall, F1-score, and precision score.

Results:
Our results showed that osteopontin and osteocalcin levels were significantly correlated with osteoporosis risk. Age, osteopontin, and osteocalcin were positively correlated with osteoporosis risk, while weight and BMI were inversely correlated. RFC had the highest performance, with an AUC of 0.86, accuracy of 84%, sensitivity of 94%, and specificity of 63.0%, indicating its superiority in predicting osteoporosis risk using a selected set of features.

Conclusions: Our machine learning-based approach achieves competitive results using only clinical features and circulating biomarker data, requiring less experimental data for training classifiers to predict osteoporosis risk. RFC is the most effective machine learning classification model for predicting osteoporosis risk in postmenopausal women using a selected set of features. This approach has the potential to assist primary care providers in identifying at-risk individuals and providing timely interventions to prevent osteoporosis-related fractures.