Risk Factor Analysis of Bone Mineral Density Based on Feature Selection in Type 2 Diabetes

Abstract

Type 2 diabetes (T2DM), one of the most common chronic diseases, predisposes bone to fragility fracture, which brings the heavy burden of medical care costs and affection on quality of life. Altered bone mineral density (BMD) is closely linked to T2DM-related bone fragility fracture. In this study, we adopt the feature selection technique to learning the most relevant or informative risk factors of BMD based on the clinical data set including general clinical data and glucose metabolic indexes of patients with T2DM. To illustrate the effectiveness and superiority of feature selection technique, eight state-of-the-art feature selection algorithms are exploited to select the subset of risk factors. This study successfully uses machine learning methods to implement risk factor analysis and prediction of BMD in patients with T2DM based on the easily obtained data in community medical institutions, which will be beneficial for the management of T2DM-related bone fracture in the primary healthcare systems.