基于神经网络的医保欺诈主动发现模型

本文阐述了运用自组织竞争型神经网络(SOM)与BP神经网络相结合,建立医保欺诈主动发现模型的原理和过程。主要介绍医疗数据的特征,海量数据初步分类和精选样本对BP神经网络进行训练的方法,最后采用遗传算法对BP神经网络的权值和阈值进行优化。研究成果较好地实现了对医保欺诈行为的主动识别。

In order to build a model for detection of fraud in medical services, the paper proposes a new algorithm by combining self-organized map neural network（SOM） with back-propagation artificial neural network（BP）.The authors firstly introduce the characteristic of medical data, and then investigate approach based on SOM to select samples for the training of BP neural network. At last, the paper applies genetic algorithm into the model, so as to optimize initial weights and biases. It turns out that the model is highly effective in intelligent recognition of fraud.