| 摘要: |
| ［摘要］ 目的 利用主成分分析和聚类分析对胰腺炎发病早期的生化指标进行研究，并依据其特征对胰腺炎病人进行分类。方法 搜集2011-05～2012-04住院的24例急性胰腺炎（AP）患者的血细胞分析、血生化检查单资料进行研究，主成分分析使用SPSS13.0软件进行分析，聚类分析使用SPSS13.0和SAS软件综合进行分析。结果 根据主成分分析结果可将胰腺炎发病早期的生化指标归为四类，分别是酶类相关变化、血液浓缩和炎症渗出状况、凝血功能及感染严重程度。聚类分析结果表明，对于多数病人来说，其发病具有规律性，以此为基础，可以为规范化治疗提供依据和参考。而对于脂肪酶（LIPA≥1 500 U/L）、淀粉酶（AMYL≥970 U/L）、谷草转氨酶(AST≥600 U/L)、谷丙转氨酶(ALT≥580 U/L)值较高以及血细胞压积（HCT≥0.58 L/L）及单核细胞(MON≥2.98×109/L)明显异常的患者，需特别注意，从而为研究胰腺炎发病机制和制定个体化治疗方案提供指导。结论 通过主成分分析和聚类分析可以充分发掘数据信息，对病人疾病状况及时作出评价，并根据其结果对病人进行分类，从而为早诊断、早治疗及预防并发症提供依据和参考。 |
| 关键词: 急性胰腺炎 主成分分析 聚类分析 特征值 贡献率 |
| DOI：10.3969/j.issn.1674-3806.2013.04.12 |
| 分类号:R 576 |
| 基金项目: |
|
| Applied research of principal component analysis and cluster analysis in the early diagnosis of pancreatitis |
|
WANG Zhi-min,WANG Gang,JI Jia-chao,et al.
|
|
Department of General Surgery,First Hospital of Shanxi Medical University,Taiyuan 030001,China
|
| Abstract: |
| ［Abstract］ Objective To use the principal component analysis and cluster analysis to study the early biochemical markers of pancreatitis，and classify the pancreatitis（AP） patients according to their characteristics.Methods The data of analysis of blood cells and blood chemistry examination in 24 AP patients admitted in hospital from May 2011 to April 2012 was investigated，the principal component analysis was performed by SPSS13.0 software，cluster analysis by SPSS13.0 and SAS software.Results According to the results of principal component analysis the early biochemical markers of AP can be classified into 4 categories，including related changes in enzymes,blood concentration and inflammatory exudation situation,coagulation and severity of infection. The cluster analysis results showed that for the majority of AP patients,its incidence had regularity，which could provide the basis and reference for standardized treatment.We should Pay particular attention to the patients with higher valves of lipase, amylase, aspartate aminotransferase, alanine aminotransferase values, hematocrit and obvious abnormalities of monocytes, then provide guidance for research of pathogenesis pancreatitis and the development of individualized treatment plan.Conclusion The data can be fully explored by principal component analysis and cluster analysis, timely evaluate the state of the patients’ disease and to classify patients based on the results, and provide the basis and reference for early diagnosis, early treatment and prevention of complications. |
| Key words: Acute pancreatitis(AP) Principal component analysis Cluster analysis Characteristic value Contribution rate |
