基于偏最小二乘-判别分析算法的癌痛规范化治疗评价

Assessment of standardized cancer pain treatment based on partial least squares discriminant analysis algorithm

目的:建立基于偏最小二乘-判别分析(PLS-DA)算法的癌痛规范化治疗评价模型,探讨医院癌痛管理的新模式。方法:提取本院2013年95名和2014年76名癌痛患者的基本信息,确定本院癌痛规范化治疗的10项评价指标,分别对患者的各项指标进行评分,构建反映患者基本信息和指标评分的多变量数据矩阵。建立PLS-DA模型,构建得分图对比研究2013和2014年的整体差异,构建载荷图筛选差异较显著的变量。结果:与2013年相比,2014年本院癌痛规范化治疗得到了改善。其中有无入院和治疗后疼痛强度评分、滴定药物种类选择和剂量换算是否正确等指标改善最为显著。结论:本研究证明了PLS-DA算法在癌痛规范化治疗评价中的有效性,可为医院癌痛管理提供新的方法。

AIM： To explore new mode of hospital cancer pain management,a model for assessing normative cancer pain treatment was established based on the partial least squares discriminant analysis（ PLS-DA） algorithm. METHODS： The basic information of 95 patients in 2013 and 76 patients in2014 of our hospital was extracted. 10 different evaluation indicators were determined for assessing the standardized cancer pain treatment and the grade of each indicator was evaluated. Based on the basic information and grade of indicator,a multivariate data matrix was built. Then,a PLS-DA model was established. The scores plot was applied to investigate the overall differences between 2013 and 2014. The variables with most significant variation were screened according to the loading plot. RESULTS： Compared with 2013,the standard degree of cancer pain treatment was improved in 2014. Several indicators,including whether there ＇s pain intensity rating at admission and before discharge,whether the drug specie used for titration and the drug dose conversion were correct showed the most significant improvement. CONCLUSION： This case study demonstrated that it was effective to employ PLS-DA algorithm to evaluate the standard degree of cancer pain treatment. This research provided a new scientific tool for the management of cancer pain in hospital.