目的 基于近红外光谱（near infrared spectrum，NIRS）技术，建立一种快速预测天舒片崩解时间的方法。方法 采集39个批次共468个样品的NIRS，对比分类和回归树（classification and regression trees，CART）算法与偏最小二乘（partial least-square，PLS）算法2种模型的预测效果，建立天舒片崩解时间预测模型。结果 经基线校正处理后建立的CART模型性能最优。与PLS模型相比该模型将相对校正均方根偏差（relative root mean square error of correction，RRMSEC）由7.43%降低至4.94%，相对预测均方根偏差（relative root mean square error of prediction，RRMSEP）由7.84%降低至7.66%。结论 NIRS技术结合CART算法预测天舒片崩解时间是可行的，为天舒片崩解时间快速无损检测提供了一种新方法。

Objective A rapid method was established to predict the disintegration time of Tianshu Tablets (天舒片) based on near infrared spectroscopy (NIRS). Methods The near-infrared spectra of 468 samples from 39 batches were collected, and the disintegration time prediction model of Tianshu tablets was established by comparing the prediction effects of the partial least squares (PLS) and classification and regression tree (CART) models. Results The performance of the CART model was the best after the spectrum was preprocessed by the baseline correction, relative root mean square error of correction (RRMSEC) value of this model was decreased from 7.43% to 4.94%, relative root mean square error of prediction (RRMSEP) value wasdecreased from 7.84% to 7.66%. Conclusion It is feasible to predict the disintegration time of Tianshu Tablets with NIR spectroscopy technology and CART algorithm, which provides a new method for rapid and non-destructive testing of the disintegration time of Tianshu tablets.

R283.6

国家“重大新药创制”科技重大专项：基于功效成分群的中药口服固体制剂先进制药与信息化技术融合示范应用（2018ZX09201010-004）