【摘要】 目的利用乳腺癌临床数据集,构建LightGBM预测模型,评价LightGBM模型预测效果。方法选取Wisconsin医学院WilliamH.Wolberg博士提供的共699例乳腺癌临床数据,构建LightGBM预测模型,并与SVM模型、XGBoost模型在准确性、灵敏度、特异性、AUC上进行对比,绘制三种模型的ROC曲线。结果经过五折交叉验证结果表明LightGBM模型在准确性上为97.14%,灵敏度为98.32%,特异性为96.54%,AUC为0.9743,在准确性、AUC方面均高于SVM模型、XGBoost模型。结论相比而言,LightGBM预测模型更具优势,分类效果更好,在利用计算机技术对乳腺癌进行辅助诊断的方面有促进作用。更多还原

【Abstract】 Objective The LightGBM prediction model was constructed based on the clinical data set of breast cancer, and the prediction effect of LightGBM model was evaluated. Methods Based on the clinical data of 699 cases of breast cancer provided by Dr. WilliamH.Wolberg of Wisconsin Medical College, the LightGBM prediction model was constructed and compared with SVM model and XGBoost model in accuracy, sensitivity, specificity and AUC. The ROC curves of the three models were drawn. Results 5-fold cross-validation showed that the accuracy, sensitivity, specificity and specificity of LightGBM model were 97.14%, 98.32%, 96.54% and 0.9743, respectively. The accuracy and AUC of AUC model were higher than those of SVM model and XGBoost model. Conclusion In contrast, LightGBM prediction model has more advantages, better classification effect, and plays a role in the use of computer technology in the auxiliary diagnosis of breast cancer.更多还原