论文部分内容阅读
目的利用支持向量机-微量元素法建立胃癌的模式识别,以期作为胃癌诊断的辅助手段。方法以血浆作为研究对象,采用电感耦合等离子体发射光谱法测定健康者与胃癌患者血清中Zn、Fe、Cu、Ni、Co、Mo、Mn、Cr和Zn/Cu 9个指标,结合支持向量机分类模式,建立胃癌诊断的模型。结果多项式核函数的平均灵敏度为93.0%、平均特异度为91.4%、平均准确度为92.6%;径向基核函数的平均灵敏度为95.0%、平均特异度为88.6%、平均准确度为93.3%。结论采用的支持向量机-微量元素法为胃癌的预防、诊断和临床治疗提供了一种简便的辅助参考方法。
Objective To establish a pattern recognition of gastric cancer using support vector machine - trace element method in order to serve as an adjunct to gastric cancer diagnosis. Methods Plasma was taken as the research object and the indexes of Zn, Fe, Cu, Ni, Co, Mo, Mn, Cr and Zn / Cu in sera of healthy and gastric cancer patients were determined by inductively coupled plasma atomic emission spectrometry. Classification model, the establishment of gastric cancer diagnosis model. Results The average sensitivity of the polynomial kernel function was 93.0%, the average specificity was 91.4% and the average accuracy was 92.6%. The average sensitivity of radial basis function was 95.0%, the average specificity was 88.6% and the average accuracy was 93.3% . Conclusions The support vector machine - trace element method provides a convenient auxiliary reference method for the prevention, diagnosis and clinical treatment of gastric cancer.