Osteoporosis is a global public health concern which is characterized by reduced bone mineral density and increased fracture risk, particularly among the elderly population. Early and accurate identification of at-risk individuals is critical for timely clinical intervention. This study investigates the application of deep learning for osteoporosis binary classification using a comprehensive clinical dataset of 1,503 patients which contains demographic information, bone mineral density measurements, serum biochemical indices, therapy records, and comorbidity history. Two deep learning architectures are employed for the classification task: a Multilayer Perceptron (MLP) network and an Autoencoder + Classifier. Model performance was assessed using 5-fold cross validation, and the MLP was superior in the classification task, achieving a test AUC of 0.848. The model found that the most influential variables for Osteoporosis classification were total proximal femur T-score and femoral neck T-score. These results are consistent with clinical diagnostic criteria for osteoporosis, and these findings demonstrate the potential of deep learning as a clinical decision support tool for osteoporosis identification.

Keywords

Deep Learning, Osteoporosis Prediction, Bioinformatics, Bone Mineral Density, Disease Prediction