Predicting the risk of hip bone fracture for older adults in Taiwan by ensemble back-propagation neural networks ensemble

Hip bone fracture is one of the most important causes of morbidity and mortality in the older adults. It is necessary to establish a prediction model to provide suggestions for elders. However, fractures are related to many factors in different groups like disease, physical performance, environmental hazards, etc. Artificial neural networks (ANNs) are suitable for this complex application. The purpose of this paper is to establish a model to predict the risk for hip fracture by ANNs. We applied ensemble method and tried 3 types of ANNs prediction model with different architectures. Finally, one of them turned out to be the best prediction model and achieved a high success rate of prediction. The area under the receiver operating characteristic (ROC) curve and the accuracy for female model are 0.91 and 0.85, and 0.99 and 0.93 for male model. This study verified the performances of ANNs to be a highly complex prediction model. Keywords: ensemble, artificial neural networks, hip fracture. !"#$%&'()* +,-./+0/ 1.2 3456789 : ; <= )>?@/ .2ABCD)+,E FG)HI J/K LMNOPQRNSTUVWXYZ[\]R^_ ) `X/@ abcde(ANN)fkl)mn opqhANN 1.234 `X)56>r:shtensemble) uvwx4ytzj#{)56|}/~/€~‚ƒ„\…†)QR‡ˆ‰ hiŠQ)56VROC‹Œ)Ž(AUC)p‘’“/”•–p‘—˜‰@™Q56) AUCšp‘’’/”•–p‘’›>klœtANNshiHI3456)Rž>