Abstract

Info

Author(s):: YU Jinjuan; LIN Yong; School of Medical Instrument and Food Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China

Keywords:: ; Keywords: osteoporotic fracture; machine learning; XGBoost algorithm; classification prediction; ten-fold crossvalidation; LASSO dimension reduction

Abstract:: Abstract: Prediction of complex diseases is an important topic in genetics research. Herein a machine learning method is introduced, taking clinical variables and genetic variables as features to predict osteoporotic fractures. After that the features of clinical phenotypes and heritable variations were selected, Logistic regression analysis and XGBoost algorithm were used to predict the characteristic variables of clinical factors, clinical factors and genetic factors. Finally, ten-fold cross validation method was used to verify the prediction results. The experimental results show that both the prediction accuracies of XGBoost and Logistic methods are improved after adding genetic factor variation as compared with using clinical factor alone. In addition, the XGBoost method is superior to Logistic regression model in the prediction of osteoporotic fractures.