Establishment And Research Of X-ray-based Osteoporosis Risk Prediction Model For Postmenopausal Women

ObjectiveOsteoporosis is a common chronic disease in postmenopausal women,with insidious onset.It can only be diagnosed when serious complications such as osteoporotic fracture occur.Early diagnosis and intervention have an important impact on delaying the occurrence and development of osteoporosis.DXA is the "gold standard" diagnostic device for osteoporosis,but due to its high cost,it is not available in all hospitals.Therefore,this study aims to construct a prediction model of osteoporosis by using clinical data and X-ray imaging data,which can predict the risk of osteoporosis without DXA instrument.MethodsA total of 156 postmenopausal women were recruited in the Affiliated Central Hospital of Shenyang Medical College from January 2020 to January 2021.The diagnostic results of dual energy X-ray absorptiometry（DXA）were regarded as the "gold standard".They were divided into osteoporotic group（n=53）and non-osteoporotic group（n=98）according to whether they had osteoporosis or not.Clinical and X-ray data of the two groups were collected.Including age,age of first menstruation,age of menopause,degree of height loss,body mass index（BMI）,number of pregnancies,previous history of lumbago,previous history of brittle fracture,cortical thickness of proximal femur,cortical index of proximal femur,cortical thickness of distal femur,cortical index of distal femur,SI,CFI,and Dorr classification.The independent risk factors of osteoporosis were screened by univariate analysis and binary logistic regression analysis,the receiver operating characteristic（ROC）curve was drawn,and the c-index was used to analyze and verify its predictive efficacy.Finally,the clinical decision curve analysis was established to provide reference for clinical decision making of medical personnel.ResultsIn univariate analysis,there were significant differences in age,degree of height reduction,number of pregnancies,proximal femoral cortical thickness,proximal femoral cortical index,distal femoral cortical thickness,distal femoral cortical index,SI,CFI and femoral Dorr classification（P < 0.05）.The results of binary logistic regression analysis showed that age,degree of height reduction Proximal femoral cortical thickness is an independent risk factor for assessing the risk of osteoporosis.The prediction model of osteoporosis is Logit（P）=-4.681+0.101*（age）+0.253*（height loss）-0.344*（proximal femur cortex thickness）.The area under the ROC curve（AUC）predicted by the prediction model was 0.827 [95% CI（0.761,0.892）];Internal verification shows that the c-index of the model is 0.823,and the model calibration curve shows that the predicted results are in good agreement with the actual results.ConclusionsIn this study,an osteoporosis risk prediction model was constructed based on three independent risk predictors: age,height reduction and proximal femoral cortical thickness.The discrimination and calibration of the model are good,which has a certain guiding value for clinical decision-making.