论文部分内容阅读
目的采用汶川大地震后一组伤员救治医疗数据,以偏最小二乘(partial least square,PLS)算法为核心,探讨影响因素并构建一种新的大规模突发公共卫生事件创伤救治预后预测模型。方法以四川省医学科学院.四川省人民医院2008年5月收治的地震伤员数据库作为建模资料来源作为训练集,建模过程包括成分提取和建立回归方程,模型优度判别采用交叉检验。运算平台采用EXCEL和Matlab软件。结果通过相关性分析得到9个因素的相关性,VIP值得到5个影响结局的重要因素,并建立了大规模突发公共卫生事件创伤救治预后预测模型。结论本研究构建的以PLS算法为基础的模型,能可靠的帮助筛选大规模公共卫生事件中创伤患者预后预测的关键性影响因素以及综合评价指标。
Objective To investigate the influencing factors and construct a new predictive model for prognosis of massive public health emergency trauma by using a set of PLS (partial least square) algorithm to treat medical data after a large earthquake in Wenchuan. . Methods The database of earthquake casualties admitted by Sichuan Provincial People’s Hospital in May 2008 was used as the training data source of the model, and the modeling process included composition extraction and regression equation establishment, and the model excellentness discriminant was cross-checked. Computing platform using EXCEL and Matlab software. Results Correlation analysis showed that 9 factors were related, VIP value obtained five important factors affecting the outcome, and established a large-scale public health emergency tracing prognosis prediction model. Conclusion The model based on PLS algorithm constructed in this study can reliably help screen key prognostic factors of traumatic patients in large-scale public health events and comprehensive evaluation index.