关于「Bergman 最小模型:远端胰岛素作用室的数学结构」,《连续血糖曲线与最小模型》给出的定量结论是什么?
结论:最小模型用两个方程把血糖与血浆胰岛素连起来:葡萄糖效能 \(S_G\) 加上远端室的胰岛素作用 \(X\),构成葡萄糖的分数消失率;\(S_I = p_3 / p_2\) 是这个消失率对胰岛素的斜率。远端室使作用滞后于胰岛素:示意注射中,血浆胰岛素在第 0 分钟最高,作用在第 38 分钟才最高,血糖在第 120 分钟降到 76.3,低于基础值。剖面计算表明,一条带噪声的曲线几乎确定不了 \(p_2\),\(p_2\) 在 0.015 到 0.060 之间时残差都在噪声底线 3.98 附近;\(S_I\) 相对稳定,\(S_G\) 与 \(p_2\) 此消彼长。下一章把模型从静脉注射换成进餐,问一个更贴近日常读数的问题:在一个人的几个参数里,哪一个最能推动他的餐后曲线。
数据口径:本书不是医疗建议,不宣称任何功效,不提供任何饮食、禁食或用药安排;任何自我试验须先咨询执业医师。 书中参数为示意;全部计算由书中程序生成。
对照数据表
下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 120, 38, 76.3
| SI = p3 / p2 = 1.250e-05 / 0.025 = 5.0e-04 | ||||
| t | insulin | G | X | SG + X |
| 0 | 88.0 | 300.0 | 0.00000 | 0.02500 |
| 5 | 61.0 | 272.5 | 0.00382 | 0.02882 |
| 10 | 45.4 | 245.0 | 0.00598 | 0.03098 |
| 20 | 31.2 | 197.0 | 0.00783 | 0.03283 |
| 30 | 26.4 | 160.5 | 0.00834 | 0.03334 |
| 60 | 21.8 | 101.8 | 0.00806 | 0.03306 |
| 90 | 17.7 | 81.6 | 0.00684 | 0.03184 |
| 120 | 14.1 | 76.3 | 0.00523 | 0.03023 |
| 180 | 10.0 | 78.8 | 0.00244 | 0.02744 |
| insulin peaks at 0 min; action X peaks at 38 min | ||||
| glucose disappearance slope Kg (10-30 min): 0.0211 per min | ||||
| check, insulin held at Ib: G(60) RK4 136.857334 exact 136.857334 |
核心公式
\[\frac{dG}{dt} = -(S_G + X)\,G + S_G\,G_b, \qquad \frac{dX}{dt} = -p_2\,X + p_3\,(I - I_b).\]
查看完整验算程序
# Chapter 3. p31: the glucose minimal model after an intravenous bolus.
# dG/dt = -(SG + X) G + SG Gb
# dX/dt = -p2 X + p3 (I(t) - Ib), SI = p3 / p2
# X is insulin action in a remote compartment (per min); plasma insulin
# I(t) is a given input, as in the 1979 model. Parameters illustrative:
# SG 0.025/min, p2 0.025/min, SI 5.0e-4 per min per microU/mL,
# Gb 90 mg/dL, Ib 8 microU/mL, G(0) 300 mg/dL. RK4, step 0.5 min.
import math
SG, P2, SI, GB, IB, G0 = 0.025, 0.025, 5.0e-4, 90.0, 8.0, 300.0
P3 = P2 * SI
def ins(t):
# 示意血浆胰岛素:快速第一相 + 缓慢第二相(microU/mL)
return (IB + 80.0 * math.exp(-t / 10.0)
+ 15.0 * (t / 40.0) * math.exp(1.0 - t / 40.0))
def rhs(t, g, x, insulin):
return (-(SG + x) * g + SG * GB, -P2 * x + P3 * (insulin(t) - IB))
def run(insulin, t_end=180.0, h=0.5):
g, x, t = G0, 0.0, 0.0
out = {0: (g, x)}
for k in range(int(round(t_end / h))):
k1 = rhs(t, g, x, insulin)
k2 = rhs(t + h / 2, g + h / 2 * k1[0], x + h / 2 * k1[1], insulin)
k3 = rhs(t + h / 2, g + h / 2 * k2[0], x + h / 2 * k2[1], insulin)
k4 = rhs(t + h, g + h * k3[0], x + h * k3[1], insulin)
g += h / 6 * (k1[0] + 2 * k2[0] + 2 * k3[0] + k4[0]) # step
x += h / 6 * (k1[1] + 2 * k2[1] + 2 * k3[1] + k4[1]) # step
t = (k + 1) * h
if abs(t - round(t)) < 1e-9:
out[int(round(t))] = (g, x)
return out
res = run(ins)
print("SI = p3 / p2 = %.3e / %.3f = %.1e" % (P3, P2, SI))
print(" t insulin G X SG + X")
for t in (0, 5, 10, 20, 30, 60, 90, 120, 180):
g, x = res[t]
print("%3d %6.1f %6.1f %.5f %.5f" % (t, ins(t), g, x, SG + x))
xmax = max(range(181), key=lambda t: res[t][1])
print("insulin peaks at 0 min; action X peaks at %d min" % xmax)
kg = (math.log(res[10][0]) - math.log(res[30][0])) / 20.0
print("glucose disappearance slope Kg (10-30 min): %.4f per min" % kg)
basal = run(lambda t: IB, 60.0)
exact = GB + (G0 - GB) * math.exp(-SG * 60.0)
print("check, insulin held at Ib: G(60) RK4 %.6f exact %.6f"
% (basal[60][0], exact))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《连续血糖曲线下面积、双室胰岛素最小模型与生酮代谢切换动力学母典》第 3 章「Bergman 最小模型:远端胰岛素作用室的数学结构」(版本 1.0.0). 5D Library. DOI 待发布
DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-286 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026286,
author = {Fan, Yafu},
title = {Areas under Continuous Glucose Curves, the Two-Compartment Minimal Model of Insulin Action and the Dynamics of the Switch to Ketone Metabolism},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}