第5章 第二节NONMEM基础模型

4,034次阅读
29 条评论

结构模型

结构模型:群体药动学模型中的结构模型是经典的药动学模型,如一级消除的一房室药动学模 型、一级吸收和一级消除的二房室模型等。结构模型的选择方法与经典药动学一致,可通 过绘制模型诊断图、统计学检验以及评估参数估算值的精密度等进行综合 评判,不能仅依赖于统计学检验。

结构模型可以根据药物的特征及数据的特点灵活地选择:
1. 如果所研究的药物的药动学特点已有文献报道,可以选用文献报道的药动学模型作为结构模型。
2. 若无现成的研究数据可以利用,可利用密集取样的数据,采用经典药动学方法拟合模型,初步判断结构模型。
3. 更为常用的方法是验证常见房室模型(一室、二室或三室等),以及米-曼氏非线性模型,还可以根据数据特征自行建立微分方程表示,再通过不同结构模型的OFV大小选择最优模型。

固定效应模型

固定效应模型:固定效应模型用于定量地考察固定效应对药动学参数的影响,模型结构包括线性、乘法、饱和、指示变量模型。比如研究不同固定效应对清除率的影响:

  1. 线性模型
    用公式表示如下:

$$
CL= θ_1 * WT
$$

  1. 乘法模型
    如果患者体重范围很大,而且药物的清除与体重相关,有时可以下式表示:

$$
CL= θ_1 * WT^{θ_2}
$$

  1. 指示变量模型
    用公式表示如下:

$$
CL=(θ_1-θ_2⋅HF)×WT
$$

  1. 饱和模型
    对于描述具有最大效应的模型很有用,比如合并使用一种能够抑制研究药物代谢的药物时,清除率可用公式表示如下

$$
CL=WT[θ_1-(θ_2 Cpss_2)/(θ_3+Cpss_2 )]
$$

随机效应模型

随机效应包括个体间变异(IIV)及残差变异,前者描述群体中个体间药动学参数的差异;而后者则描述了观测值的无法解释的变异。除此之外,近年在药动学研究中还引入了其他变异,如不同用药阶段的变异(inter-occasional variability, IOV)及不同研究间的变异(inter-study variability, ISV)。e
$$
药动学参数的个体间变异以随机变量η表示,其均值为0,而方差ω2,表示为η=N(0, ω2);\
残差误差,即观察值与预测值之差, 以随机变量ε表示,其均值为0,而方差σ2,表示为ε=N(0, σ2)。
$$
Θ θ:西塔 Theta

Ε η:依塔 Eta
Ω ω:欧米伽 Omega

Ε ε:艾普西龙 Epsilon
∑ σ:西格玛容 Sigma

几种模型结构表示
加法模型:

$$
CL_j=CL+η{j}^{CL}\
C
{ij}^0=C_{ij}+ε_{ij}\
$$

其中CLj为某一个体参数真值,CL为参数群体典型值, C_{ij}^0为浓度观察值, C_{ij}为浓度的模型预测值。 、eij分别为个体间及随机变异。

比例模型:

$$
CL_j=CL(1+η{j}^{CL})\
C
{ij}^0=C_{ij}(1+ε_{ij})\
$$

指数模型:

$$
CL_j=CL×EXP(η_{j}^{CL})\
$$

混合模型:

$$
C_{ij}^0=C_{ij}^P⋅ε_1+ε_2
$$

不同用药阶段变异模型(IOV)

$$
θi=θμ×exp⁡( η_i+η_1⋅OCC_1+η_2⋅OCC_2+…+η_O⋅OCC_O)
$$

$ERROR
  IPRED=F
  W=THETA(X)                                     加和型
  W=THETA(Y)*IPRED                               比例型
  W=SQRT(THETA(X)**2)+(THETA(Y)*IPRED)**2)       混合型

上式中OCC为指示变量,例如处于第一治疗阶段,则OCC1值为1,其他则为0。这种方法的不足是对于某一患者,若其接受过多个治疗阶段,得到的药动学参数可能会有多个

参数的估算精度

NONMEN可输出THETA、OMEGA和SIGMA参数估计值的标准 误差(standarderror,SE)。与相应的参数估算值比较,可计算获得参数估计值的相对精度———相对标准误差百分比(RSE 一般而言,固定效应的参数相对精度低于30

示例:

$PROB infusion_PK                    ;建模描述
$INPUT C ID TIME TAD NMT CONC=DV AMT RATE MDV DOSE GROUP SEX AGE WEIGHT HEIGHT BMI  ;数据结构
$DATA ..\infusion_3cp.csv IGNORE=C   ;数据路径,导入以及忽略数据
$SUBROUTINES ADVAN11 TRANS4          ;模型结构,三室
$PK                                  ;$PK模块
   TVCL=THETA(1)                     ;CL典型值
   CL=TVCL*EXP(ETA(1))               ;CL个体值,固定效应模型为指数模型(个体间变异)

   TVV1=THETA(2)                     ;V1典型值
   V1=TVV1*EXP(ETA(2))               ;V1个体值,固定效应模型为指数模型

   TVV2=THETA(3)
   V2=TVV2*EXP(ETA(3))

   TVQ2=THETA(4)
   Q2=TVQ2*EXP(ETA(4))

   TVV3=THETA(5)
   V3=TVV3*EXP(ETA(5))

   TVQ3=THETA(6)
   Q3=TVQ3*EXP(ETA(6))

   S1=V1/1000                        ;系数sur=1000
   S2=V2/1000
   S3=V3/1000

$ERROR                               ;$ERROR 定义残差模型(个体内变异)
  IPRED=F
  Y=F*(1+ERR(1))+ERR(2)              ;混合残差模型(个体内变异)

$THETA                               ;$THETA 初值设定
 (0,52.7)     ; [CL] 1               ;THETA1初值
 (0,10)      ; [V1] 2                ;THETA2初值
 (0,15)     ; [V2] 3                 ;THETA3初值
 (0,40)     ; [Q2] 4                 ;THETA4初值
 (0,20)     ; [V3]                   ;THETA5初值
 (0,20)     ; [Q3]                   ;THETA6初值

$OMEGA                               ;$OMEGA 初值设定
0.02   ;  ETA1[CL]                   ;$OMEGA1 初值设定
0.02   ;  ETA2[V1]                   ;$OMEGA3 初值设定
0.02   ;  ETA3[V2]                   ;$OMEGA3 初值设定
0 FIX  ;  EATA4[Q2]                  ;$OMEGA4 初值设定
0.02   ;  EATA5[V3]                  ;$OMEGA5 初值设定
0 FIX  ;  EATA6[Q3]                  ;$OMEGA6 初值设定

$SIGMA                               ;$SIGMA初值设定
0.02  ;  EPS[PROP]                   ;比例变异系数
0.1   ;   EPS[ADD]                   ;加和变异系数

;$SIM(234567) ONLYSIM NSUB=1         ; 种子数
$ESTIMATION METHOD=COND INTER MAXEVAL=9999 PRINT=1 SIG=3  ; 估计方法 
$COVARIANCE PRINT=E                  ; 协方差矩阵

$TABLE ID TIME NMT TAD DOSE PRED IPRED CWRES ONEHEADER NOPRINT FILE=SDTAB1  ; 输出内容和格式
正文完
 
评论(29 条评论)
2025-12-10 05:15:18 回复

I don’t think the title of your article matches the content lol. Just kidding, mainly because I had some doubts after reading the article.

 Windows  Edge  中国广西玉林市电信
2025-12-22 05:59:11 回复

Your point of view caught my eye and was very interesting. Thanks. I have a question for you. https://www.binance.info/en-IN/register-person?ref=A80YTPZ1

 Windows  Edge  中国广西玉林市电信
2026-01-10 23:26:50 回复

Thank you for your sharing. I am worried that I lack creative ideas. It is your article that makes me full of hope. Thank you. But, I have a question, can you help me? https://accounts.binance.com/si-LK/register-person?ref=LBF8F65G

 Windows  Edge  中国广西玉林市电信
2026-01-23 11:45:32 回复

我经常计划, 能像你们一样多旅行。很开心。 傳統村落 有教育意义的 旅游页面! 你们真棒!

 Windows  Chrome  乌克兰

Thanks for sharing. I read many of your blog posts, cool, your blog is very good. https://accounts.binance.info/ru/register?ref=O9XES6KU

 Windows  Edge  中国广西玉林市电信
2026-02-06 20:04:03 回复

Your point of view caught my eye and was very interesting. Thanks. I have a question for you.

 Windows  Edge  中国广西玉林市电信
2026-02-13 04:36:20 回复

Can you be more specific about the content of your article? After reading it, I still have some doubts. Hope you can help me.

 Windows  Edge  中国广西玉林市电信
2026-02-23 23:32:09 回复

Your article helped me a lot, is there any more related content? Thanks! https://www.binance.com/ar-BH/register?ref=S5H7X3LP

 Windows  Edge  中国广西玉林市电信
2026-03-10 11:23:27 回复

I don’t think the title of your article matches the content lol. Just kidding, mainly because I had some doubts after reading the article.

 Windows  Edge  中国广西玉林市电信
2026-03-21 18:11:43 回复

Can you be more specific about the content of your article? After reading it, I still have some doubts. Hope you can help me. https://www.binance.com/register?ref=IHJUI7TF

 Windows  Edge  中国广西玉林市电信
2026-03-25 11:43:14 回复

Your point of view caught my eye and was very interesting. Thanks. I have a question for you. https://www.binance.info/register?ref=IHJUI7TF

 Windows  Edge  中国广西百色市电信
2026-04-08 17:54:58 回复

Can you be more specific about the content of your article? After reading it, I still have some doubts. Hope you can help me. https://www.binance.info/register?ref=IXBIAFVY

 Windows  Edge  中国广西钦州市电信
2026-04-11 09:48:27 回复

Your point of view caught my eye and was very interesting. Thanks. I have a question for you.

 Windows  Edge  中国广西玉林市电信
2026-04-13 17:50:19 回复

Thank you for your sharing. I am worried that I lack creative ideas. It is your article that makes me full of hope. Thank you. But, I have a question, can you help me?

 Windows  Edge  中国广西百色市电信
2026-04-16 01:43:06 回复

Your point of view caught my eye and was very interesting. Thanks. I have a question for you.

 Windows  Edge  中国广西钦州市电信
2026-04-17 04:57:24 回复

Your article helped me a lot, is there any more related content? Thanks!

 Windows  Edge  中国广西钦州市电信
2026-04-23 04:50:40 回复

Thank you for your sharing. I am worried that I lack creative ideas. It is your article that makes me full of hope. Thank you. But, I have a question, can you help me?

 Windows  Edge  中国广西钦州市电信
2026-04-28 23:10:03 回复

Your point of view caught my eye and was very interesting. Thanks. I have a question for you.

 Windows  Edge  中国广西玉林市电信
2026-04-29 11:57:11 回复

Thanks for sharing. I read many of your blog posts, cool, your blog is very good. https://www.binance.com/register?ref=IHJUI7TF

 Windows  Edge  中国广西玉林市电信
2026-05-02 01:21:18 回复

Thanks for sharing. I read many of your blog posts, cool, your blog is very good. https://accounts.binance.bh/kz/register?ref=K8NFKJBQ

 Windows  Edge  中国广西玉林市电信
2026-05-03 18:35:42 回复

Can you be more specific about the content of your article? After reading it, I still have some doubts. Hope you can help me. https://accounts.binance.bh/id/register/person?ref=UM6SMJM3

 Windows  Edge  中国广西钦州市电信
2026-05-07 04:30:03 回复

Can you be more specific about the content of your article? After reading it, I still have some doubts. Hope you can help me.

 Windows  Edge  中国广西玉林市电信
2026-05-13 14:57:02 回复

Thank you for your sharing. I am worried that I lack creative ideas. It is your article that makes me full of hope. Thank you. But, I have a question, can you help me?

 Windows  Edge  中国广西玉林市电信
2026-05-21 06:21:11 回复

Can you be more specific about the content of your article? After reading it, I still have some doubts. Hope you can help me. https://www.binance.info/register?ref=JW3W4Y3A

 Windows  Edge  中国广西玉林市电信
2026-05-22 04:52:45 回复

Can you be more specific about the content of your article? After reading it, I still have some doubts. Hope you can help me.

 Windows  Edge  中国广西玉林市电信
2026-06-02 05:13:53 回复

Your article helped me a lot, is there any more related content? Thanks!

 Windows  Edge  中国广西钦州市电信
2026-06-11 10:42:27 回复

Your point of view caught my eye and was very interesting. Thanks. I have a question for you.

 Windows  Edge  中国广西钦州市电信
2026-06-14 13:27:57 回复

Can you be more specific about the content of your article? After reading it, I still have some doubts. Hope you can help me.

 Windows  Edge  中国广西玉林市电信
2026-06-14 18:19:56 回复

Your point of view caught my eye and was very interesting. Thanks. I have a question for you.

 Windows  Edge  中国广西玉林市电信