将逆传播（ＢＰ）网络应用于癌症的元素分析诊断中，根据人发中２２个微量元素浓度的测定，建立了有相应的２２个输入节点的三层ＢＰ网络，对癌症进行判别分析。详细考察了各种参数组合对预报结果的影响，对参数的选择进行了优化，５次交叉性检验所得到的预报准确率为７９２％±３４２％。结果表明，该方法可作为医学诊断的辅助手段。 Based on the determination of 22 trace elements in human hair, a BP network with 22 corresponding input nodes was established to diagnose the cancer. The effects of various parameters on the prediction results are examined in detail. The selection of parameters is optimized. The accuracy of the prediction obtained by the five cross tests is 79.2% ± 3.42%. The results show that this method can be used as an adjunct to medical diagnosis.