| 摘 要: 广义 Youden指数突破了传统识别精度统计技术局限,具有广泛的应用前景。为了进一步发展 Youden指数识别精度统计技术,提出了广义 Youden指数参数估计方法及密度比模型下半参数估计方法,并通过仿真分析与已有的核密度估计方法进行对比。仿真结果显示,大样本下该半参数估计方法提升置信区间覆盖率约3%,缩短长度约50%,证明了其合理性和高效性。最后,把广义 Youden指数半参数估计方法应用在筛选高识别性能乳腺癌生物标志物的实际应用中,验证了它的实际应用价值 |
| 关键词: 广义 ROC 广义 Youden指数 密度比模型 半参数估计 |
|
中图分类号: TP399
文献标识码: A
|
| 基金项目: 国家自然科学基金资助项目(12471304) |
|
| Estimation and Application of Generalized Youden Index |
|
HE Jiali, YANG Jianping
|
(School of Science, Zhejiang Sc-i Tech University, Hangzhou 310018, China)
galilly@163.com; yangjp@zstu.edu.cn
|
| Abstract: The Generalized Youden Index overcomes the limitations of traditional statistical techniques for recognition accuracy and exhibits broad application prospects. To further advance the Youden Index-based statistical techniques for recognition accuracy, this study proposes parameter estimation method for the Generalized Youden Index and a semiparametric estimation approach under the density ratio model. A simulation analysis was conducted to compare these methods with the existing kernel density estimation method. The simulation results indicate that, in large samples, the semiparametric method improves the confidence interval coverage rate by approximately 3% and reduces its length by about 50% , demonstrating its rationality and efficiency. Finally, the semiparametric estimation method for the Generalized Youden Index was applied to the practical task of screening high-accuracy breast cancer biomarkers,
highlighting its practical value. |
| Keywords: generalized ROC generalized Youden index density ratio model semiparametric estimation |
