Risk of Bone Loss Among Male Patients With Abnormal Glucose Metabolism Based on Machine Learning

Context: Associations between abnormal glucose metabolism and osteoporosis remain uncertain. What’s more, osteoporosis is usually considered as a health problem for women, and its influence and harm to men are often ignored. Objective: The objective of the study was to develop a osteoporosis or osteopenia incidence risk nomogram in China’s male patients with abnormal glucose metabolism by machine learning methods. Design, Setting, and Patients: This was a cross-sectional survey, which was part of a longitudinal (REACTION) study. The data came from a chronic disease database conducted in Ningde City and Wuyishan City, Fujian Province, China from June 2011 to January 2012. In the end, 1758 male patients with abnormal glucose metabolism were included for data analysis. Main Outcomes Measures: The 67 related variables are obtained by questionnaires, physical examinations, blood tests and auxiliary examinations. Results: The predictors in the prediction model included age, body mass index, alanine transaminase, fracture history, smoking history, fruit flavored drink, strong physical activity, and nap time. The model displayed moderate predictive power with a C-index of 0.653 (95%CI: 0.622-0.684) and an area under the ROC curve of 0.653. In verification set, the C-index reached 0.691 (95%CI: 0.636-0.746). The risk threshold was 8%-100% according to the analysis of the decision curve analysis, and the nomogram could be applied in clinical practice. Conclusions: This nomogram incorporating 8 features can be used to predict osteoporosis or osteopenia incidence risk in male patients with abnormal glucose metabolism.