目的系统评价基于极端梯度提升(eXtreme Gradient Boosting,XGBoost)算法构建的重症加强护理病房(Intensive Care Unit,ICU)死亡风险预测模型的研究现况。方法检索知网、万方、维普、PubMed、Embase、Web of Science、Scopus数据库,搜...目的系统评价基于极端梯度提升(eXtreme Gradient Boosting,XGBoost)算法构建的重症加强护理病房(Intensive Care Unit,ICU)死亡风险预测模型的研究现况。方法检索知网、万方、维普、PubMed、Embase、Web of Science、Scopus数据库,搜集有关基于XGBoost算法构建的ICU死亡风险预测模型的研究,检索时限均为建库至2023年2月18日。由2名研究者独立筛选文献,提取资料并评价纳入研究的偏倚风险后,进行定性系统评价。结果共纳入12篇文献,纳入模型的受试者工作特征曲线下面积为0.750~0.941。10篇文献适用性较好,其余2篇文献适用性不清楚。12篇文献均存在高偏倚风险,偏倚主要来自于不合适的研究数据来源、研究对象的纳排标准不清晰、预测因子定义与评估不一致、基于单因素分析法筛选预测因子、缺乏完善的模型性能评估等。结论现有基于XGBoost算法构建的ICU死亡风险预测模型具有较好的区分度,但其临床预测的准确性还尚不明确。未来还需进一步完善相关研究设计,避免研究中的各类偏倚风险,加强模型的外部验证,确保模型在临床实践中的可行性及有效性。展开更多
The machining unit of hobbing machine tool accounts for a large portion of the energy consumption during the operating phase.The optimization design is a practical means of energy saving and can reduce energy consumpt...The machining unit of hobbing machine tool accounts for a large portion of the energy consumption during the operating phase.The optimization design is a practical means of energy saving and can reduce energy consumption essentially.However,this issue has rarely been discussed in depth in previous research.A comprehensive function of energy consumption of the machining unit is built to address this problem.Surrogate models are established by using effective fitting methods.An integrated optimization model for reducing tool displacement and energy consumption is developed on the basis of the energy consumption function and surrogate models,and the parameters of the motor and structure are considered simultaneously.Results show that the energy consumption and tool displacement of the machining unit are reduced,indicating that energy saving is achieved and the machining accuracy is guaranteed.The influence of optimization variables on the objectives is analyzed to inform the design.展开更多
针对机床的机械故障频发且装配因素难以识别的问题,提出了基于贝叶斯网络的机床装配情景异常推理识别方法。以机械零部件多尺度运动分析为切入点,建立了机床功能-元动作的多尺度映射模型,利用故障模式及影响分析(Failure Mode and Effec...针对机床的机械故障频发且装配因素难以识别的问题,提出了基于贝叶斯网络的机床装配情景异常推理识别方法。以机械零部件多尺度运动分析为切入点,建立了机床功能-元动作的多尺度映射模型,利用故障模式及影响分析(Failure Mode and Effect Analysis,FMEA)方法建立了机床元动作单元关键装配情景构成模型。基于装配情景构成模型建立了元动作单元装配情景的贝叶斯网络结构,利用证据推理法实现了元动作单元装配情景异常概率的智能推理。以蜗轮转动元动作单元为例,构建了蜗轮转动单元装配情景初始贝叶斯网络,获取了蜗轮转动元动作输出的异常概率(由装配因素引起)为2.35%;以蜗轮转动故障为起点进行了贝叶斯网络反向推理,识别出导致蜗轮转动故障的各装配情景异常概率。元动作装配情景的异常识别为实现机床故障装配因素的追溯提供理论依据。展开更多
文摘目的系统评价基于极端梯度提升(eXtreme Gradient Boosting,XGBoost)算法构建的重症加强护理病房(Intensive Care Unit,ICU)死亡风险预测模型的研究现况。方法检索知网、万方、维普、PubMed、Embase、Web of Science、Scopus数据库,搜集有关基于XGBoost算法构建的ICU死亡风险预测模型的研究,检索时限均为建库至2023年2月18日。由2名研究者独立筛选文献,提取资料并评价纳入研究的偏倚风险后,进行定性系统评价。结果共纳入12篇文献,纳入模型的受试者工作特征曲线下面积为0.750~0.941。10篇文献适用性较好,其余2篇文献适用性不清楚。12篇文献均存在高偏倚风险,偏倚主要来自于不合适的研究数据来源、研究对象的纳排标准不清晰、预测因子定义与评估不一致、基于单因素分析法筛选预测因子、缺乏完善的模型性能评估等。结论现有基于XGBoost算法构建的ICU死亡风险预测模型具有较好的区分度,但其临床预测的准确性还尚不明确。未来还需进一步完善相关研究设计,避免研究中的各类偏倚风险,加强模型的外部验证,确保模型在临床实践中的可行性及有效性。
基金This work was supported in part by the National Natural Science Foundation of China(Grant Nos.51975075 and 52105506)the Chongqing Technology Innovation and Application Program,China(Grant No.cstc2020jscx-msxmX0221).
文摘The machining unit of hobbing machine tool accounts for a large portion of the energy consumption during the operating phase.The optimization design is a practical means of energy saving and can reduce energy consumption essentially.However,this issue has rarely been discussed in depth in previous research.A comprehensive function of energy consumption of the machining unit is built to address this problem.Surrogate models are established by using effective fitting methods.An integrated optimization model for reducing tool displacement and energy consumption is developed on the basis of the energy consumption function and surrogate models,and the parameters of the motor and structure are considered simultaneously.Results show that the energy consumption and tool displacement of the machining unit are reduced,indicating that energy saving is achieved and the machining accuracy is guaranteed.The influence of optimization variables on the objectives is analyzed to inform the design.
文摘针对机床的机械故障频发且装配因素难以识别的问题,提出了基于贝叶斯网络的机床装配情景异常推理识别方法。以机械零部件多尺度运动分析为切入点,建立了机床功能-元动作的多尺度映射模型,利用故障模式及影响分析(Failure Mode and Effect Analysis,FMEA)方法建立了机床元动作单元关键装配情景构成模型。基于装配情景构成模型建立了元动作单元装配情景的贝叶斯网络结构,利用证据推理法实现了元动作单元装配情景异常概率的智能推理。以蜗轮转动元动作单元为例,构建了蜗轮转动单元装配情景初始贝叶斯网络,获取了蜗轮转动元动作输出的异常概率(由装配因素引起)为2.35%;以蜗轮转动故障为起点进行了贝叶斯网络反向推理,识别出导致蜗轮转动故障的各装配情景异常概率。元动作装配情景的异常识别为实现机床故障装配因素的追溯提供理论依据。
