论文部分内容阅读
目的通过建立误差反向传播人工神经网络(Back Popagation Artificial Neural Network,BP-ANN)预测模型分析本地区与胃部高危疾病有关的影响因素并评价模型预测效果。方法于2012年10月至2014年11月,通过现场调查的方式对719名徐州市三级甲等医院消化科门诊患者进行问卷调查,并且收集其胃镜病理及相关资料。应用Logistic回归模型对影响胃部高危疾病发生的饮食、生活状况及习惯、身体状况等因素进行单因素和多因素分析;运用训练集资料建立BP神经网络,通过测试集142例调查对象评价该方法在实际应用中的预测精度。结果对调查数据进行胃部高危疾病的Logistic回归单因素分析,分析显示女性、年龄、胃癌家族史、喜食烧烤类食物、生活节奏快等10个因素有统计学意义(P<0.05);多因素分析显示,吸烟、饮用自来水、喜食豆制品、常生气、年龄和生活节奏快与胃部高危疾病有关。BP神经网络筛查的效果较好。结论建立好的BP神经网络可以用于筛检胃癌高危人群。
OBJECTIVE: To analyze the influencing factors of high risk disease in the stomach and to evaluate the predictive effect of the model by establishing a back propagation artificial neural network (BP-ANN) prediction model. Methods From October 2012 to November 2014, 719 outpatients from the gastroenterology department of Grade A Hospitals of Xuzhou City were surveyed by means of on-the-spot investigation and their pathological features of gastroscopy and related materials were collected. Logistic regression model was used to analyze single factor and multifactorial factors such as diet, living conditions and habits, body condition and other factors that affect the occurrence of gastric high risk disease. BP neural network was established by using training set data and evaluated by 142 test subjects Prediction accuracy in practical applications. Results Univariate analysis of logistic regression analysis showed that there were statistically significant (P <0.05) 10 factors such as female, age, family history of gastric cancer, eating grilled food and fast rhythm of life. Factor analysis showed that smoking, drinking tap water, eating soy products, often angry, age and fast-paced life-related high-risk stomach disease. BP neural network screening better. Conclusion The established BP neural network can be used to screen high-risk gastric cancer patients.