一种基于概率校正和集成学习的肠癌肝转移预测模型

提出一种基于概率校正和集成学习的机器学习模型,用来预测患者肠癌肝转移的概率。首先将AdaBoost和Class-bal-anced SVM的概率结果进行校正,再将其结果和Logistic回归的预测结果进行集成,获得最终的预测结果。预测模型在复旦大学附属肿瘤医院的肠癌患者数据集上与其他算法如AdaBoost、Class-balanced SVM、Logistic回归算法进行了比较,结果显示该模型具有更好的AUC性能,更适合于医生的临床辅助诊断。模型的AUC性能在UCI数据集上进一步得到了验证。

A PROBABILITY CALIBRATION AND ENSEMBLE LEARNING BASED COLORECTAL CANCER LIVER METASTASIS PREDICTION MODEL

A probability calibration and ensemble learning based machine learning model is proposed to predict the probability of colorectal cancer liver metastasis.At first the AdaBoost and Class-balanced SVM probability results are calibrated;then those results are assembled with Logistic regression prediction results to produce the final prediction result.The prediction model is compared against other algorithms such as AdaBoost,Class-balanced SVM and Logistic regression on the colorectal cancer patient data set of...