Based on drug sensitivity data and extensive gene expression data, a model was constructed by multivariate analysis with the partial least squares method type 1. Further, the model was optimized using modeling power and genetic algorithm. Thereby, the degree of contribution of the respective genes to drug sensitivity was determined to select genes with a high degree of contribution. In addition, the levels of gene expression in specimens were analyzed, and then the drug sensitivity was predicted based on the model. The predicted values agreed well with those drug sensitivity values determined experimentally. The drug sensitivity-predicting method provided by the present invention enables assessment of the effectiveness of a drug prior to administration using small quantities of specimens associated with diseases such as cancer. Since this enables the selection of the most suitable drug for each patient, the present invention is very useful in improving a patient's quality of life (QOL).
根据药物敏感性数据和大量
基因表达数据,通过偏最小二乘法 1 型多变量分析构建了一个模型。此外,还利用建模能力和遗传算法对模型进行了优化。据此,确定了各
基因对药物敏感性的贡献程度,以选择贡献程度高的
基因。此外,还分析了标本中
基因的表达
水平,然后根据模型预测了药物敏感性。预测值与实验测定的药物敏感性值非常吻合。通过本发明提供的药物敏感性预测方法,可以在用药前使用与癌症等疾病相关的少量标本来评估药物的有效性。这样就能为每位患者选择最合适的药物,因此本发明对提高患者的生活质量(QOL)非常有用。