Chinese men should have a higher prostate-specific antigen （PSA） ＂gray zone＂ than the traditional value of 2.5-10.0 ng ml-1 since the incidence of prostate cancer （PCa） in Chinese men is relative low. We hypothe...Chinese men should have a higher prostate-specific antigen （PSA） ＂gray zone＂ than the traditional value of 2.5-10.0 ng ml-1 since the incidence of prostate cancer （PCa） in Chinese men is relative low. We hypothesized that PSA density （PSAD） could improve the rate of PCa detection in Chinese men with a PSA higher than the traditional PSA ＂gray zone.＂ A total of 461 men with a PSA between 2.5 and 20.0 ng ml-1, who had undergone prostatic biopsy at two Chinese centers were included in the analysis. The men were then further divided into groups with a PSA between 2.5-10.0 ng ml-1 and 10.1-20.0 ng ml-1. Receiver operating characteristic （ROC） curve was used to evaluate the efficacy of PSA and PSAD for the diagnosis of PCa. In men with a PSA of 2.5-10.0 ng ml-1 or 10.1-20.0 ng ml-z, the areas under the ROC curve were higher for PSAD than for PSA. This was consistent across both centers and the cohort overall. When the entire cohort was considered, the optimal PSAD cut-off for predicting PCa in men with a PSA of 2.5-10.0 ng m1-1 was 0.15 ng ml-2 ml-2, with a sensitivity of 64.4% and specificity of 64.6%. The optimal cut-off for PSAD in men with a PSA of 10.1-20.0 ng m1-1 was 0.33 ng ml-1 ml-1, with a sensitivity of 60.3% and specificity of 82.7%. PSAD can improve the effectiveness for PCa detection in Chinese men with a PSA of 2.5-10.0 ng ml-1 （traditional Western PSA ＂gray zone＂） and 10.1-20.0 ng ml-2 （Chinese PSA ＂gray zone＂）.展开更多
The random forests （RF） algorithm, which combines the predictions from an ensemble of random trees, has achieved significant improvements in terms of classification accuracy. In many real-world applications, however...The random forests （RF） algorithm, which combines the predictions from an ensemble of random trees, has achieved significant improvements in terms of classification accuracy. In many real-world applications, however, ranking is often required in order to make optimal decisions. Thus, we focus our attention on the ranking performance of RF in this paper. Our experi- mental results based on the entire 36 UC Irvine Machine Learning Repository （UCI） data sets published on the main website of Weka platform show that RF doesn＇t perform well in ranking, and is even about the same as a single C4.4 tree. This fact raises the question of whether several improvements to RF can scale up its ranking performance. To answer this question, we single out an improved random forests （IRF） algorithm. Instead of the information gain measure and the maximum-likelihood estimate, the average gain measure and the similarity- weighted estimate are used in IRF. Our experiments show that IRF significantly outperforms all the other algorithms used to compare in terms of ranking while maintains the high classification accuracy characterizing RF.展开更多
文摘Chinese men should have a higher prostate-specific antigen （PSA） ＂gray zone＂ than the traditional value of 2.5-10.0 ng ml-1 since the incidence of prostate cancer （PCa） in Chinese men is relative low. We hypothesized that PSA density （PSAD） could improve the rate of PCa detection in Chinese men with a PSA higher than the traditional PSA ＂gray zone.＂ A total of 461 men with a PSA between 2.5 and 20.0 ng ml-1, who had undergone prostatic biopsy at two Chinese centers were included in the analysis. The men were then further divided into groups with a PSA between 2.5-10.0 ng ml-1 and 10.1-20.0 ng ml-1. Receiver operating characteristic （ROC） curve was used to evaluate the efficacy of PSA and PSAD for the diagnosis of PCa. In men with a PSA of 2.5-10.0 ng ml-1 or 10.1-20.0 ng ml-z, the areas under the ROC curve were higher for PSAD than for PSA. This was consistent across both centers and the cohort overall. When the entire cohort was considered, the optimal PSAD cut-off for predicting PCa in men with a PSA of 2.5-10.0 ng m1-1 was 0.15 ng ml-2 ml-2, with a sensitivity of 64.4% and specificity of 64.6%. The optimal cut-off for PSAD in men with a PSA of 10.1-20.0 ng m1-1 was 0.33 ng ml-1 ml-1, with a sensitivity of 60.3% and specificity of 82.7%. PSAD can improve the effectiveness for PCa detection in Chinese men with a PSA of 2.5-10.0 ng ml-1 （traditional Western PSA ＂gray zone＂） and 10.1-20.0 ng ml-2 （Chinese PSA ＂gray zone＂）.
文摘The random forests （RF） algorithm, which combines the predictions from an ensemble of random trees, has achieved significant improvements in terms of classification accuracy. In many real-world applications, however, ranking is often required in order to make optimal decisions. Thus, we focus our attention on the ranking performance of RF in this paper. Our experi- mental results based on the entire 36 UC Irvine Machine Learning Repository （UCI） data sets published on the main website of Weka platform show that RF doesn＇t perform well in ranking, and is even about the same as a single C4.4 tree. This fact raises the question of whether several improvements to RF can scale up its ranking performance. To answer this question, we single out an improved random forests （IRF） algorithm. Instead of the information gain measure and the maximum-likelihood estimate, the average gain measure and the similarity- weighted estimate are used in IRF. Our experiments show that IRF significantly outperforms all the other algorithms used to compare in terms of ranking while maintains the high classification accuracy characterizing RF.
