Biomarker mining method and system

The invention discloses a biomarker mining method and system, and the method comprises the steps: preliminarily determining differential proteins through the fusion screening of statistical learning and an XGBoost feature selection technology; carrying out secondary screening by utilizing an SVM (Support Vector Machine) algorithm based on Shapley feature weight to obtain proteins with high classification criticality as biomarker candidates; and completing the verification of the biomarker by combining a target correlation scoring algorithm of the mapping knowledge domain with a DISEASES algorithm. According to the method, the biomarker mining speed and accuracy are remarkably improved, the mining process is simplified, and the reliability and correlation verification of the biomarker are enhanced. 本发明公开了一种生物标志物挖掘方法及系统，包括：通过统计学习和XGBoost特征选择技术的融合筛选，初步确定差异蛋白质；利用基于Shapley特征权重的SVM算法进行二次筛选，得到极具分类关键性的蛋白质作为生物标志物候选；通过知识图谱的靶标相关性评分算法结合DISEASES算法，完成生物标志物的验证。本发明显著提高了生物标志物挖掘的速度和准确性，简化了挖掘过程，增强了生物标志物的可信度和相关性验证。