高级检索

应用BP神经网络预测前列腺癌流行趋势

Using BP Network to Predict the Prevalence of Prostate Cancer

  • 摘要: 为评价BP(Back Propagation)神经网络模型在疾病预测中的应用价值,并为前列腺癌的人群防治工作提供依据,应用BP神经网络模型对武汉市居民前列腺癌流行趋势进行预测.以1990~1995年资料建立模型预测1996~1997年流行水平和趋势,结果发病率回代平均误差率和R2分别为38.2%和0.7734,外推预测平均误差率为51.5%;死亡率回代平均误差率和R2分别为48.9%和0.9515,外推预测平均误差率为25.1%.以1990~1997年资料建立模型预测1998~2000年流行趋势,结果发病率回代平均误差率和R2分别为3.24%和0.8195;死亡率回代平均误差率和R2分别为4.02%和0.9554.外推预测结果显示:今后几年内,武汉市居民前列腺癌的发病率呈相对稳定状态,死亡率呈缓慢上升趋势.

     

    Abstract: To evaluate the applying value of BP network in disease prediction,to give clues for population prevention of prostate cancer.The prevalence of prostate cancer of WuHan residents(male)was predicted using BP network.Using the data of the year 1990~ 1995 to predict the prevalence of prostate in 1996~1997,the results showed that:as the incidence of the prostate cancer,the fitting average error was 3.83%,R2 is 0.7734,and the predicting average error was 5.15%.As the mortality rate of the prostate cancer,the fitting average error was 4189%,R2 was 019515,and the predicting average error is 2.51%.Using the data of the year 1990~1997 to predict the prevalence rate of 1998~2000,the results showed that:As the incidence of the prostate cancer,the fitting average error was 3124%,R2 was 018195.As the mortality rate of the prostate cancer,the fitting average error was 4.02%,R2 was 019554.The predicting results showed that:in the next few years,the incidence of prostate cancer will be a relatively steady state and the mortality rate of prostate cancer will show a slowly rising trend.

     

/

返回文章
返回