云模型优化径向基函数神经网络算法研究

Research on Cloud Model Optimization Radial Basis Function Neural Network Algorithm

径向基函数(RBF)神经网络广泛用于各类医学预测模型中,针对RBF神经网络隐含层高斯径向基函数的参数确定困难,影响癌症预后模型的因素具有多样性和模糊性等问题。利用云模型优化RBF神经网络算法,通过高维云变换确定RBF隐含层神经元,优化RBF神经网络结构。用来自美国国家癌症研究所监测、流行病学和最终结果(SEER)数据库的4771例食管鳞状细胞癌患者数据建模仿真与传统的仿真对比,证明该模型预测生存期的C-index为0.705,远高于肿瘤等级、列线图和RBF神经网络(0.598、0.627和0.632),能更好更准确地对食管鳞状细胞癌患者进行预后预测。

Radial basis function(RBF)neural network is widely used in various medical prediction models.It is difficult to determine the parameters of Gaussian radial basis function in the hidden layer of RBF neural network,and it cannot solve the problems such as diversity and fuzziness of factors affecting cancer prognosis models.The cloud model was used to optimize the RBF neural network algorithm,and the RBF hidden layer neurons were determined by high-dimensional cloud transformation,so as to optimize the structure of the RBF neural network.Finally,the comparison between the modeling simulation data of 4771 patients with esophageal squamous cell carcinoma(ESCC)from the surveillance,epidemiology and end results(SEER)database of the American national cancer institute and the traditional simulation proved that the C-index of survival prediction of the model was 0.705,which was much higher than that of the tumor grade,nomogram and RBF neural network(0.598,0.627 and 0.632).The prognosis of patients with ESCC could be better and more accurately predicted.