体外诊断试剂盒线性范围研究线性区间评价资料及可报告区间建立

体外诊断试剂盒线性范围研究/线性区间评价资料及可报告区间建立,,,,,,,,,

实验名称,试剂盒线性范围研究/线性区间评价及可报告区间建立,,,,,,,,

实验目的,确定试剂盒的线性范围/线性区间可报告区间上限,,,,,,,,

实验方法,"依据体外诊断试剂分析性能评估系列指导原则(征求意见稿)以及体外诊断检测系统性能评价方法第4部分：线性区间与可报告区间

基本要求

（1）样本基质应与临床实验样本相似，但不可采用含有对测定方法具有明确干扰作用物质的样本，如溶血、脂血、黄疸或含有某些特定药物的样本。进行血清学标志物检测时，理想的样本为分析物浓度接近预期测定上限的混合人血清。

（2）建立一种定量测定方法的线性范围时，需在预期测定范围内选择7-11个浓度水平。如将预期测定范围加宽至130%，在此范围内选择更多的浓度水平，然后依据实验结果逐渐减少数据点直至表现出线性关系，可发现最宽的线性范围。

（3）当对标称线性参数进行验证时，需在已知线性范围内选择5-7个浓度水平。

（4）无论是建立或验证线性范围，所选用的浓度水平应可覆盖整个预期测定范围并包括与临床有关的重要评价浓度，如最小测定浓度或线性范围的最低限、不同的医学决定水平、最大测定浓度或线性范围的高限等。",,,,,,,,

说明,公式已在excel中设定好，在红色区域录入数据即可自动计算出结果，最终结论的得出需三批数据。,,,,,,,,

标本配制：,,,,,,,,,

11个浓度水平的样本制备,,,,,,,"建立线性范围：需测定9-11个浓度水平，每个浓度水平重复测定3-4次。

验证标称线性参数：需测定4-6个浓度水平，每个浓度水平重复测定3-4次。

所有样本应在一次运行中或几次间隔很短的运行中随机测定，最好在一天之内完成。",,

样本号,1,2,3,4,5,6,,,

低浓度血清(ml),1.00,0.90,0.80,0.70,0.60,0.50,,,

高浓度血清(ml),0.00,0.10,0.20,0.30,0.40,0.50,,,

样本号,7,8,9,10,11,,,,

低浓度血清(ml),0.40,0.30,0.20,0.10,0.00,,,,

高浓度血清(ml),0.60,0.70,0.80,0.90,1.00,,,,

5个浓度水平的样本制备,,,,,,,,,

样本号,1,2,3,4,5,,,,

低浓度血清(ml),1.00,0.75,0.50,0.25,0.00,,,,

高浓度血清(ml),0.00,0.25,0.50,0.75,1.00,,,,

检测结果：,,,,,,,,,

样本号,理论浓度,测定Y1,测定Y2,测定Y3,测定Y4,Y均值,Y标准差s,t1,t2

1,25.00,25.08,24.90,25.01,25.00,25.00,0.07,1.10,-1.33

2,42.50,42.48,42.49,42.60,42.59,42.54,0.06,0.90,-0.91

3,60.00,60.00,60.03,59.96,60.07,60.02,0.05,1.19,-1.15

4,77.50,77.48,77.59,77.44,77.56,77.52,0.07,1.02,-1.09

5,95.00,94.98,94.92,94.97,95.03,94.98,0.04,1.24,-1.19

6,112.50,112.47,112.42,112.59,112.45,112.48,0.07,1.45,-0.83

7,130.00,130.06,129.96,129.98,130.02,130.01,0.05,1.23,-1.08

8,147.50,147.44,147.52,147.58,147.49,147.51,0.06,1.24,-1.13

9,165.00,165.02,165.06,164.97,165.04,165.02,0.04,0.87,-1.40

10,182.50,182.48,182.48,182.54,182.59,182.52,0.05,1.30,-0.82

11,200.00,200.05,199.98,199.99,199.95,199.99,0.04,1.37,-1.02

t值临界值,,,,,,,,,

样本数(测定次数),显著性水平,,,,样本数(测定次数),显著性水平,,,

,0.05,0.025,0.01,0.005,,0.05,0.025,0.01,0.005

3,1.153,1.155,1.155,1.155,4,1.463,1.481,1.492,1.496

如t1、t2值大于临界值，则相应的可疑值为离群值。,,,,,,,,,

"进行多项回归分析

对数据组进行多项回归分析，得到一级、二级与三级多项式。一级多项式为直线，二级多项式表示上升曲线或下降曲线，三级多项式表示S形曲线（在测量范围两端具有明显的非线性）。

多项式方程如下：

级数多项式回归自由度（Rdf）

一级Y=b0+b1X2

二级Y=b0+b1X+b2X23

三级Y=b0+b1X+b2X2+b3X34

3.5对回归方程进行线性检验

多元回归方程中以bi表示的系数为回归系数。在二级与三级方程中，b2与b3为非线性系数。对回归方程进行线性检验就是对每个非线性系数作t检验，判断回归系数与零是否有显著性差异。b0与b1不反映非线性，故不需对其进行检验。对b2与b3的检验方法如下：

计算统计量t，计算公式为：t=bi/SEi

其中，SEi为每个非线性系数的斜率标准误，计算公式为：

其中，Y为回归方程预测值，与为测定均值。",,,,,,,,,

,,,,,,,,,,

回归分析结果,,,样品数：,11,测定次数,4,,,

回归方程,方程因子,方程因子值,SEi,t-test(t=bi/Sei）,标准估计误差Sy.x,自由度df=L•R-Rdf,最适方程,结果分析,

"一次方程

(y=b1X+b0)",b0,-0.01,0.01,-0.82,,,,,

,b1,1.00,0.00,9002.81,0.02,42,,,

二次方程(y=b2X2+b1X+b0),b0,-0.02,0.03,-0.91,,,,,

,b1,1.00,0.00,1838.56,,,,t0.05(ν)=,2.020

,b2,0.00,0.00,-0.57,0.02,41,否,小于，无统计学意义,

三次方程(y=b3X3+b2X2+b1X+b0),b0,-0.04,0.05,-0.80,,,,,

,b1,1.00,0.00,549.10,,,,,

,b2,0.00,0.00,-0.46,,,,t0.05(ν)=,2.021

,b3,0.00,0.00,0.39,0.02,40,否,小于，无统计学意义,

,,,,,,,,,,

"临床标准的线性检验,如回归分析满足要求，则可省略",,,,,,,,,

如ADL＜5%，则拟合的多项式无临床意义,,,,,,,,,

样本号,测量均值,一次方程预期值,二次方程预期值,三次方程预期值,二次与一次绝对偏倚,二次与一次相对偏倚%,三次与一次绝对偏倚,三次与一次相对偏倚%,

1,25.00,24.99,24.98,24.98,-0.01,-0.025%,-0.01,-0.041%,

2,42.54,42.53,42.53,42.53,0.00,-0.006%,0.00,-0.004%,

3,60.02,60.01,60.01,60.01,0.00,0.001%,0.00,0.006%,

4,77.52,77.51,77.51,77.51,0.00,0.003%,0.01,0.007%,

5,94.98,94.97,94.97,94.97,0.00,0.004%,0.01,0.006%,

6,112.48,112.47,112.48,112.48,0.00,0.004%,0.00,0.004%,

7,130.01,130.00,130.00,130.00,0.00,0.003%,0.00,0.001%,

8,147.51,147.50,147.50,147.50,0.00,0.002%,0.00,0.000%,

9,165.02,165.02,165.02,165.02,0.00,0.000%,0.00,-0.002%,

10,182.52,182.52,182.52,182.52,0.00,-0.001%,0.00,-0.002%,

11,199.99,199.99,199.98,199.98,-0.01,-0.003%,0.00,-0.001%,

从上表数据判定绝对偏倚和相对偏倚是否满足要求（法定或自定义标准）。,,,,,,,,,

可报告区间建立（本标准仅针对上限，下限为定量限）,,,,,,,,,

高值样本号,理论浓度,样本稀释比例,重复测量结果1,重复测量结果2,重复测量结果3,实测均值,还原浓度,相对偏倚,

样本1,2000,5,387.6,387.6,416,397.0666667,1985.333333,-0.73%,

,,25,75.352,79.384,76.752,77.16266667,1929.066667,-3.55%,

,,50,37.552,43.348,38.308,39.736,1986.8,-0.66%,

,,100,22.002,18.856,21.342,20.73333333,2073.333333,3.67%,

,,200,8.739,8.818,8.689,8.748666667,1749.733333,-12.51%,

样本2,1800,5,366.12,359.1,362.52,362.58,1812.9,0.72%,

,,25,72.1512,71.748,70.236,71.3784,1784.46,-0.86%,

,,50,35.676,36.486,36.4212,36.1944,1809.72,0.54%,

,,100,16.9506,19.8018,19.1484,18.6336,1863.36,3.52%,

,,200,8.0413,7.9766,7.7254,7.914433333,1582.886667,-12.06%,

样本3,1600,5,309.28,311.68,324.32,315.0933333,1575.466667,-1.53%,

,,25,62.4768,61.5808,66.1504,63.40266667,1585.066667,-0.93%,

,,50,34.0448,33.7568,34.8224,34.208,1710.4,6.90%,

,,100,16.8624,14.7152,14.7328,15.4368,1543.68,-3.52%,

,,200,6.5512,7.0936,7.1144,6.919733333,1383.946667,-13.50%,

以相对偏倚不大于设定偏倚为判断标准，选取三份样本的相对偏倚均不大于设定偏倚的最大稀释倍数为方法推荐的最大稀释倍数（小于此倍数的三份样本所有稀释比例的相对偏倚均不大于设定偏倚），测量区间上限与最大稀释倍数的乘积为该方法可报告区间的上限。,,,,,,,,,

假定10%为偏倚上限，则3份样本在稀释倍数100倍及以下时，实测均值与理论值的偏倚均小于±10%，而200倍时两份样本偏倚已超过允许偏倚，故本项目最大稀释倍数为100。,,,,,,,,,

ADL计算：,,,,,,,,,

内容,,,二次方程,三次方程,,,,,

指导原则方法ADL=,,,0.008%,0.006%,,,,,

行标方法ADL中间计算值,,,0.0006,0.0009,,,,,

行标方法ADL=,,,0.003%,0.004%,,,,,

非线性临界值PctBnd=,,,5%,5%,,,,,

行标公式,,,,指导原则公式,,,,,

,,,,,,,,,,

,,,,,,,,,,

,,,,,,,,,,

,,,,,,,,,,

,,,,,p(x):最优拟合二阶或三阶方程的拟合值,,,,

,,,,,b0+b1x：拟合一阶方程的拟合值,,,,

对数据进行精密度检验，线性评价前提,,,,,L：样本数,,,,

,,,,,¯c=（y1+y2+y3+……+yn）/n,,,,

指导原则公式,,,行标公式,,n：样本数乘以重复次数,,,,

,,,,,d:最优拟合方程的阶,,,,

,,,,,σ：最优拟合发能给出的回归标准误,,,,

,,,,,Rdf:回归自由度,,,,

,,,,,R：重复测量的次数,,,,

,,,,,C：常数,,,,

,,,,,,,,,,

样本号,（yi-（Y一阶回归））^2,（yi-（Y二阶回归））^2,（yi-（Y三阶回归））^2,SDr计算中间值,CVr计算中间值,,,,

1,0.0000,0.0096,0.0061,0.0165,0.0000,,,,

2,0.0123,0.0021,0.0016,0.0119,0.0000,,,,

3,0.0076,0.0001,0.0005,0.0072,0.0000,,,,

4,0.0139,0.0011,0.0051,0.0136,0.0000,,,,

5,0.0057,0.0001,0.0024,0.0055,0.0000,,,,

6,0.0161,0.0001,0.0034,0.0159,0.0000,,