基于集成学习的骨质疏松性骨折预测研究

Prediction of osteoporotic fracture based on ensemble learning

骨质疏松性骨折是老年人发病和死亡的重要原因之一,建立高效的预测模型为老年人尽早提供诊断和治疗建议十分必要。实验利用Stacking构建了一种异构分类器EtDtb-S,将16个相关性较高的特征作为特征向量,选用极端随机树(ET)、基于决策树的装袋集成模型(DTB)作为初级学习器,逻辑回归作为次级学习器进行集成。实验验证将EtDtb-S与单模型、同构分类器进行骨质疏松性骨折预测对比,结果表明异构分类器相对于最优单模型预测精度提高2.8%,相对于最优同构分类器预测精度提高1.5%,具有更高的预测性能。

Osteoporotic fracture is one of the important causes of morbidity and death in the elderly.It is necessary to establish an efficient predictive model to provide diagnosis and treatment suggestions for the elderly as soon as possible.In the experiment,Stacking is used to construct a heterogeneous classifier EtDtb-S which uses 16 highly-correlated features as feature vectors,and selects extreme random trees and decision tree-based bagging ensemble models as primary learners,and logistic regression as the secondary learner for ensemble learning.Experimental verification compares EtDtb-S with single model and isomorphic classifiers for osteoporotic fracture prediction.The results show that the prediction accuracy of the heterogeneous classifier is increased by 2.8%and 1.5%as compared with the optimal single model and the optimal isomorphic classifier,respectively.The proposed method has better prediction of osteoporotic fracture.