肾移植受者的环孢素剂量预测

Prediction of dosage of ciclosporin in kidney transplant recipients

目的:对肾移植患者的环孢素给药剂量进行前瞻性的预测。方法:收集本院的肾移植病例随机分成2组,其中2/3的数据作为训练组样本,1/3的数据作为测试组样本,运用人工神经网络技术建立环孢素给药剂量预测模型,并用小样本的预测对模型进行验证。结果:完成环孢素给药剂量预测模型的构建,小样本(10个病例,共30次剂量)的预测结果表明,与实际给药剂量相比,预测给药剂量误差小于10%的有27次,误差在10%～20%之间的有3次。结论:人工神经网络可用于环孢素给药剂量预测。

OBJECTIVE This study was performed to predict dosage of ciclosporin in renal transplant recipients.METHODS Cases undergone kidney transplantation were collected. Among them, two third was assigned as training samples and one third was assigned as testing samples. Predictive model of dosage of ciclosporin was developed based on artificial neural nel- work. After that,small samples were forecasted using the predictive model. RESULTS The predictive model was established. Predictive dosages of ciclosporin from 10 patients （30 points in all） demonstrated that the deviation of 27 points was less than 10%,3 points was 10 to 20%, compared to actual dosages. CONCLUSION Dosage of ciclosporin may be forecasted using artifi cial neural network.