基于LM算法的神经网络在冠心病诊断中的应用

为了解决冠心病诊断的BP神经网络存在收敛速度慢、容易陷入局部极小以及常出现误诊断等问题．提出一种基于LM算法改进的神经网络诊断系统，包括样本信息选取、病情信息量化、网络学习训练和诊断等过程。临床实验应用表明，这种诊断系统不仅具有算法稳健、样本拟合精度高等优点，而且其诊断效果优于BP算法。

The Coronary Heart Disease Diagnosis by Neural Network Based on Levenberg-Marquardt Algorithm

There are some problems in the coronary heart disease diagnosis system based on BP neural network, such as the low convergent rate, easy local minimum in network training, frequent errors in diagnosis, and so on. To solve above problems, an improved neural network diagnosis system based on Levenberg-Marquardt (LM) algorithm is presented, which includes the main process of sample selection, patients' information quantification, network training and diagnosis. The application of clinic experimentation shows the diagnosis system not only possesses the merits of algorithm stability and high precision in sample fitting, but also has a superior diagnosis effect to that of BP Algorithm.