Hubei Provincial Center for Disease Control and Prevention, 6 Zhuodaoquan North Road, Wuhan, 430079, Hubei, China.
Outpatient Department, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, 430030, Hubei, China.
Parasitol Res. 2022 Aug;121(8):2457-2460. doi: 10.1007/s00436-022-07583-8. Epub 2022 Jun 29.
Predictive models for prognosis of small sample advanced schistosomiasis patients have not been well studied. We aimed to construct prognostic predictive models of small sample advanced schistosomiasis patients using two machine learning algorithms, k nearest neighbour (kNN) and support vector machine (SVM) utilising routinely available data under the government medical assistance programme. The predictive models were derived from 229 patients from Xiantao and externally validated by 77 patients of Jiayu, two county-level cities in Hubei province, China. Candidate predictors were selected according to expert opinions and literature reports, including clinical features, sociodemographic characteristics, and medical examinations results. An area under the receiver operating characteristic curve (AUC), sensitivity, and specificity were used to evaluate the models' predictive performances. The AUC values were 0.879 for the kNN model and 0.890 for the SVM model in the training set, 0.852 for the kNN model, and 0.785 for the SVM model in the external validation set. The kNN and SVM models can be used to improve the health services provided by healthcare planners, clinicians, and policymakers.
针对小样本晚期血吸虫病患者预后的预测模型尚未得到很好的研究。本研究旨在使用两种机器学习算法（k 最近邻算法（kNN）和支持向量机（SVM）），利用政府医疗救助计划下的常规可用数据，为小样本晚期血吸虫病患者构建预后预测模型。该预测模型来源于中国湖北省仙桃市的 229 名患者，并通过该市下属的嘉鱼县的 77 名患者进行外部验证。候选预测因子根据专家意见和文献报道进行选择，包括临床特征、社会人口统计学特征和体检结果。受试者工作特征曲线下面积（AUC）、敏感性和特异性用于评估模型的预测性能。在训练集中，kNN 模型和 SVM 模型的 AUC 值分别为 0.879 和 0.890，kNN 模型和 SVM 模型在外部验证集中的 AUC 值分别为 0.852 和 0.785。kNN 和 SVM 模型可用于改善医疗保健规划者、临床医生和政策制定者提供的卫生服务。
