目的　为克服经典的最小模型技术 (MMT)样本数太多 (3 0个 )、历时太长 (3小时 )的缺陷 ,以减少样本数所获得的信息对估计参数和重建胰岛素曲线的误差降至最小范围。方法　试对经典 MMT进行样本数减至 12个和 10个、时间缩至 2小时改良 ,分别用 3 0、12和 10个样本三种方案的胰岛素改良 MMT(IMMMT)评估 16例葡萄糖耐量正常 (NGT)者和 13例 2型糖尿病病人的胰岛素敏感性 (SI)和葡萄糖效应 (SG)。结果　不仅 12个样本 ,而且 10个样本的 SI和 SG与 3 0个样本的 SI和 SG在每个组值都相近 ,相关性高 (r=0 74～ 0 92 ,P<0 0 0 0 1)。结论　提示样本数减至 10～ 12个、试验时间缩至 2小时改良 IMMMT是可行的。 The purpose is to overcome the shortcomings of the classic minimum model technology (MMT) with too many samples (30) and durations that are too long (3 hours) to reduce the error in estimating parameters and reconstructing the insulin curve by reducing the number of samples obtained. The minimum range. Methods The number of samples in the classic MMT was reduced to 12 and 10, and the time was reduced to 2 hours. The improved insulin-modified MMT (IMMMT) with 30, 12 and 10 samples was used to evaluate 16 cases of normal glucose tolerance ( Insulin Sensitivity (SI) and Glucose Effect (SG) in 13 patients with type 2 diabetes and NGT). The results were not only 12 samples, but SI and SG of 10 samples and SI and SG of 30 samples were similar in each group, with high correlation (r=0 74 to 0 92, P<0 0 0 0 1 ). Conclusion It is feasible to improve the IMMMT by reducing the number of samples to 10 to 12 and reducing the test time to 2 hours.