基于最小二乘回归的定量指标一致性评价的样本量公式研制
Development of Sample Size Formulas for Least Square Regression-Based Consistency Evaluation of Quantitative Indicators
陈飞龙 1于淼 1徐涛1
作者信息
- 1. 中国医学科学院基础医学研究所 北京协和医学院基础学院流行病与卫生统计学系, 北京 100005
- 折叠
摘要
目的 构建并验证基于最小二乘回归法的定量资料一致性评价的样本量计算公式.方法 根据最小二乘回归法定量一致性评价的原理,结合统计推断和公式推导,获得基于回归常数和回归系数的样本量计算公式.同时,使用3 个实例数据验证公式的准确性,并与Bland-Altman(BA)法样本量计算公式结果进行对比.结果 通过公式推导得到回归法定量一致性评价的样本量计算公式,并用实例验证了本公式的准确性.此外,本公式与BA法样本量公式所得结果具有一定差异,在回归法样本量公式计算结果下,回归分析与BA法分析均能得到一致性结论;而在BA法公式结果下,回归分析与BA法分析一致性结论有时不成立.结论 本研究首次提出用于回归法定量一致性评价的样本量公式,为该领域研究提供了方法学支持.
Abstract
Objective To develop and verify the sample size formulas for quantitative data consistency e-valuation based on the least square regression method.Methods According to the principle of least square regres-sion-based quantitative consistency evaluation,statistical inference,and formula derivation,we developed the for-mulas for calculating sample size based on regression constant and regression coefficient.Furthermore,the accuracy of the formulas was verified by the data of three examples,and the results were compared with those of the sample size formula established based on the Bland-Altman(BA)method.Results The sample size formulas for regres-sion-based quantitative consistency evaluation were deduced,and the accuracy of the formulas was verified by three examples.In addition,the results obtained with this formula had differences compared with those of the sample size formula established based on the BA method.Furthermore,consistent conclusions could be obtained by regression analysis and BA analysis with the sample size calculated with the regression method.However,with the sample size calculated based on the BA method,the consistency conclusion of regression analysis and BA analysis was some-times not valid.Conclusion A sample size formula for quantitative consistency evaluation based on the regression method was proposed for the first time,which provided methodological support for the research in this field.
关键词
最小二乘回归/一致性评价/样本量/公式/定量指标Key words
least square regression/consistency evaluation/sample size/formula/quantitative indicator引用本文复制引用
基金项目
首都卫生发展科研专项(首发2024-2G-4211)
中央高水平医院临床科研专项(2022-PUMCH-C-058)
出版年
2024
国家科技期刊平台
NSTL
万方数据