基于BP神经网络的中毒诊断模型的设计和建立

目的：考察基于BP神经网络的中毒诊断模型的诊断效果。方法：首先运用粗糙集算法对41例已确诊的中毒病例进行属性约简，然后将其分成训练集和预示集，通过对训练集的学习建立中毒诊断模型，对预示集进行中毒诊断。结果：预示集病例的诊断正确率为87．5％。结论：该模型设计合理，可以辅助医生通过中毒者体征对中毒进行初步诊断。

A Diagnosis Model of Poisoning Based on BP Neural Network

Objective： To observe the effect of the diagnosis medel of poisoning based on BP neural network. Methods：41 poisoning cases were selected to be the original data, after the attributes reduction with the rough set algorithm, and the cases were divided into training set and test set. Learning from the training set, the diagnosis medel based on BP neural network was established. Results： The model was used to diagnose the test set, and the correct rate was 87.5 %. Conclusion： The medel can aid doctors or experts in forensic medicine in the practice of poisoning diagnosis and is proved to be reasonable.