目的 以核磁共振（NMR）分析不同年龄SD大鼠尿样中内源性代谢物的改变实验数据为基础，提出新的积分方法。方法 通过在原有固定步长积分的基础上引入步长变动区间，使数据积分区间可以根据峰的位置进行一定范围的调整，形成变步长积分方法。以固定步长和变步长积分方法，对实际实验数据进行比较研究。结果 变步长积分方法既能够增强样品聚类能力，能够减少差异代谢物指认缺失现象的发生。结论 变步长积分方法克服了固定步长积分方法存在的不足，解决了固定步长积分方法不能够同时兼顾图谱分辨率和减少由于环境引起的化学位移变化的矛盾。

Objective To present a new integrating algorithm for NMR analysis based on the data of endogenous metabolites in urine which changed by the age of SD rats. Methods This new algorithm—variable step integrating algorithm (VSIA) is by adding variable step range on fixed step, thus the integrating range can be adjusted by different peak widths. The integral method of VSIA included four procedures such as denoising of sample data, recognition of peak-valley point, domain integral based on input parameters, and output data of integral domain. In this paper, we compared VSIA and fixed step algorithm, using pattern recognitions, principal component analysis (PCA), and partial least squares-discriminate analysis (PLS-DA), to analyze the integral data from the NMR spectra. Results VSIA can significantly enhance the aggregative of different group samples, and can also reduce the loss of different metabolites caused by naive data preprocessing. By integrating effectively according to the signal peak position, VSIA can both enhance the spectrum resolution and reduce the chemical shift changes caused by environment, thus the loss of fixed step algorithm could be made up. Conclusion This study suggests that VSIA could be applied to metabonomic studies, and also could be extended to the other multi- dimensional data processing analysis.