论文部分内容阅读
目的:利用BP人工神经网络建模研究1英里跑(1 600 m跑)与实验室直接测试之间的数学关系,以构建估测大学生最大摄氧量的模型。方法:第1阶段选择80名大学生(男女各40人,年龄19~23岁)进行实验室直接测试作为建模标准,建立预测模型;第2阶段选择63名大学生(男39人,女24人,年龄19~23岁)测试进行模型验证,根据室外估测方程Jog方程,利用Neuroph Studio进行BP人工神经网络建模。结果:根据测试者年龄、性别、体重、1 600 m跑时间和跑后即刻心率5个变量作为输入变量构建BP人工神经网络,网络结构为5-9-1,采用第1阶段80%样本反复训练建模、20%样本验证模型输出的错误率并成功控制在千分之一,进一步采用第2阶段测试数据验证发现,神经网络模型估测与实验室测试值的相关系数r为0.923,P<0.01,高于Jog方程的0.895,且神经网络模型估测与实验室测试值没有显著性差异,t=0.06,P>0.05,通过Bland-Altman分析显示人工神经网络模型的偏倚程度为-8.3~8.4,比Jog方程理想。结论:人工神经网络具有快速、有效、精确和便于实际应用等特点,是解决这些没有精确数学方程的最大摄氧量估测问题的非常好途径。
OBJECTIVE: To study the mathematical relationship between 1 mile run (1 600 m run) and laboratory direct test using BP artificial neural network modeling to build a model to estimate the maximum oxygen uptake of undergraduates. Methods: In Phase 1, 80 college students (40 males and 40 females, aged 19 to 23 years old) were selected to conduct direct laboratory tests as modeling standards to establish a prediction model. The second stage consisted of 63 college students (39 males and 24 females , Age 19 to 23 years) test model validation, according to outdoor estimation equation Jog equation, the use of Neuroph Studio BP artificial neural network modeling. Results: According to the age, sex, weight, 1 600 m running time and heart rate immediately after running as input variables, a BP artificial neural network was constructed. The network structure was 5-9-1. The first stage 80% Training model, 20% of the sample to verify the model output error rate and the successful control of the one-thousandth, the further use of the second phase of the test data validation found that neural network model and laboratory testing the correlation coefficient r was 0.923, P <0.01, which is higher than 0.895 of Jog’s equation, and there is no significant difference between neural network model estimation and laboratory test values, t = 0.06, P> 0.05. Bland-Altman analysis shows that the artificial neural network model has a bias of -8.3 ~ 8.4, better than the Jog equation. Conclusion: Artificial neural network is fast, effective, accurate and practical. It is a very good way to solve the problem of maximum oxygen uptake without these exact math equations.