PDF(Pubmed)
Abstract：
Viral diseases are a severe public health issue worldwide. During the coronavirus pandemic, the use of alcohol-based sanitizers was recommended by WHO. Enveloped viruses are sensitive to ethanol, whereas non-enveloped viruses are considerably less sensitive. However, no quantitative analysis has been conducted to determine virus ethanol sensitivity and the important variables influencing the inactivation of viruses to ethanol. This study aimed to determine viruses\' sensitivity to ethanol and the most important variables influencing the inactivation of viruses exposed to ethanol based on machine learning. We examined 37 peer-reviewed articles through a systematic search. Quantitative analysis was employed using a decision tree and random forest algorithms. Based on the decision tree, enveloped viruses required around ≥ 35% ethanol with an average contact time of at least 1 min, which reduced the average viral load by 4 log10. In non-enveloped viruses with and without organic matter, ≥ 77.50% and ≥ 65% ethanol with an extended contact time of ≥ 2 min were required for a 4 log10 viral reduction, respectively. Important variables were assessed using a random forest based on the percentage increases in mean square error (%IncMSE) and node purity (%IncNodePurity). Ethanol concentration was a more important variable with a higher %IncMSE and %IncNodePurity than contact time for the inactivation of enveloped and non-enveloped viruses with the available organic matter. Because specific guidelines for virus inactivation by ethanol are lacking, data analysis using machine learning is essential to gain insight from certain datasets. We provide new knowledge for determining guideline values related to the selection of ethanol concentration and contact time that effectively inactivate viruses.
摘要：
病毒性疾病是世界范围内严重的公共卫生问题。在冠状病毒大流行期间，世卫组织建议使用含酒精消毒剂.包膜病毒对乙醇敏感,而无包膜病毒的敏感性相当低。然而,没有进行定量分析来确定病毒乙醇的敏感性和影响病毒对乙醇灭活的重要变量。这项研究旨在确定病毒对乙醇的敏感性，以及基于机器学习的影响暴露于乙醇的病毒灭活的最重要变量。我们通过系统搜索检查了37篇同行评审的文章。使用决策树和随机森林算法进行定量分析。根据决策树,包膜病毒需要≥35％的乙醇，平均接触时间至少为1分钟，这将平均病毒载量降低了4log10。在有和没有有机物质的无包膜病毒中，≥77.50％和≥65％乙醇，延长接触时间≥2分钟是4log10病毒减少所必需的，分别。基于均方误差(%IncMSE)和节点纯度(%IncNodePurity)的增加百分比,使用随机森林评估重要变量。对于用可用的有机物质灭活包膜和无包膜病毒,乙醇浓度是比接触时间更重要的变量,具有更高的%IncMSE和%IncNodePurity。因为缺乏乙醇灭活病毒的具体指南，使用机器学习进行数据分析对于从某些数据集中获得洞察力至关重要。我们为确定与选择有效灭活病毒的乙醇浓度和接触时间有关的指导值提供了新知识。
