Physics-informed neural networks(PINNs)have become an attractive machine learning framework for obtaining solutions to partial differential equations(PDEs).PINNs embed initial,boundary,and PDE constraints into the los...Physics-informed neural networks(PINNs)have become an attractive machine learning framework for obtaining solutions to partial differential equations(PDEs).PINNs embed initial,boundary,and PDE constraints into the loss function.The performance of PINNs is generally affected by both training and sampling.Specifically,training methods focus on how to overcome the training difficulties caused by the special PDE residual loss of PINNs,and sampling methods are concerned with the location and distribution of the sampling points upon which evaluations of PDE residual loss are accomplished.However,a common problem among these original PINNs is that they omit special temporal information utilization during the training or sampling stages when dealing with an important PDE category,namely,time-dependent PDEs,where temporal information plays a key role in the algorithms used.There is one method,called Causal PINN,that considers temporal causality at the training level but not special temporal utilization at the sampling level.Incorporating temporal knowledge into sampling remains to be studied.To fill this gap,we propose a novel temporal causality-based adaptive sampling method that dynamically determines the sampling ratio according to both PDE residual and temporal causality.By designing a sampling ratio determined by both residual loss and temporal causality to control the number and location of sampled points in each temporal sub-domain,we provide a practical solution by incorporating temporal information into sampling.Numerical experiments of several nonlinear time-dependent PDEs,including the Cahn–Hilliard,Korteweg–de Vries,Allen–Cahn and wave equations,show that our proposed sampling method can improve the performance.We demonstrate that using such a relatively simple sampling method can improve prediction performance by up to two orders of magnitude compared with the results from other methods,especially when points are limited.展开更多
An improved safety analysis based on the causality diagram for the complex system of micro aero-engines is presented.The study is examined by using the causality diagram in analytical failure cases due to rupture or p...An improved safety analysis based on the causality diagram for the complex system of micro aero-engines is presented.The study is examined by using the causality diagram in analytical failure cases due to rupture or pentration in the receiver of micro turbojet engine casing,and the comparisons are also made with the results from the traditional fault tree analysis.Experimental results show two main advantages：（1）Quantitative analysis which is more reliable for the failure analysis in jet engines can be produced by the causality diagram analysis;（2）Graphical representation of causality diagram is easier to apply in real test cases and more effective for the safety assessment.展开更多
背景:膝骨关节炎是一种常见的关节软骨及周围组织损伤的慢性炎症性疾病,而免疫细胞在膝骨关节炎免疫炎症反应中起到重要作用,但其中的具体机制仍有待深入研究。目的:采用孟德尔随机化方法来评估731种免疫细胞表型与膝骨关节炎风险之间...背景:膝骨关节炎是一种常见的关节软骨及周围组织损伤的慢性炎症性疾病,而免疫细胞在膝骨关节炎免疫炎症反应中起到重要作用,但其中的具体机制仍有待深入研究。目的:采用孟德尔随机化方法来评估731种免疫细胞表型与膝骨关节炎风险之间的潜在因果关系。方法:使用全基因组关联分析(GWAS)目录中公开获取731种免疫细胞表型的全基因组关联分析统计数据(从GCST0001391到GCST0002121)和IEUGWAS数据库中膝骨关节炎的全基因组关联分析数据(ebi-a-GCST007090)。采用逆方差加权法、MR-Egger回归法、加权中位数法、加权模型法和简单模型法来研究免疫细胞与膝骨关节炎之间的因果关系。敏感性分析用于检验孟德尔随机化分析结果是否可靠,然后以同样方法进行反向孟德尔随机化分析。结果与结论:①正向分析结果表明,共有4种免疫细胞表型与膝骨关节炎有显著的因果关系(FDR<0.20),其中B细胞中的CD27 on CD24+CD27+(OR=1.026,P=0.00026,Pfdr=0.18)、髓系细胞中的CD33 on CD33dim HLA DR-(OR=1.014,P=0.00050,Pfdr=0.18)以及Treg细胞中的CD45RA+CD28-CD8br%CD8br(OR=1.001,P=0.00078,Pfdr=0.18)与膝骨关节炎风险呈直接的正向因果关联;单核细胞中PDL-1 on monocyte(OR=0.952,P=0.00098,Pfdr=0.18)与膝骨关节炎风险呈直接的负向因果关联。②反向分析结果表明,当膝骨关节炎作为暴露数据时,与731种免疫细胞表型均不具有显著因果关系(FDR<0.20)。③敏感性分析结果显示:双向孟德尔随机化的Cochran’s Q检验和MR-Egger回归法结果P值均大于0.05,表明免疫细胞表型与膝骨关节炎之间的因果效应分析不存在显著的异质性和多效性。④上述结果证实,CD27 on CD24+CD27+,CD33 on CD33dim HLA DR-,CD45RA+CD28-CD8br%CD8br以及PDL-1 on monocyte免疫细胞表型与膝骨关节炎之间可能具有较为显著的潜在因果关系,这为研究膝骨关节炎的生物学机制及探索膝骨关节炎的早期防治提供有价值的线索,也为干预性药物的开发提供了新的方向。展开更多
背景:骨质疏松性骨折是骨质疏松症最严重的并发症,既往的研究已经证实了肠道菌群对骨骼组织具有调节作用,肠道菌群与骨质疏松性骨折有着重要关系,但是二者之间的因果关系尚不清楚。目的:使用孟德尔随机化(MR)方法探索肠道菌群与骨质疏...背景:骨质疏松性骨折是骨质疏松症最严重的并发症,既往的研究已经证实了肠道菌群对骨骼组织具有调节作用,肠道菌群与骨质疏松性骨折有着重要关系,但是二者之间的因果关系尚不清楚。目的:使用孟德尔随机化(MR)方法探索肠道菌群与骨质疏松性骨折之间的因果关系。方法:从IEU Open GWAS数据库和芬兰数据库R9中分别获得了肠道菌群和骨质疏松性骨折的GWAS数据集,以肠道菌群作为暴露因素,骨质疏松性骨折作为结局变量,采用随机效应逆方差加权法、MR-Egger回归、加权中位数法、简单模型法以及加权模型法进行孟德尔随机化分析来评估肠道菌群与骨质疏松性骨折之间是否存在因果关系,通过敏感性分析来检验结果的可靠性和稳健性,并进行反向孟德尔随机化分析来进一步验证正向孟德尔随机化分析中确定的因果关系。结果与结论:①此孟德尔随机化分析结果表明,肠道菌群与骨质疏松性骨折之间存在因果关系。放线菌目(OR=1.562,95%CI:1.027-2.375,P=0.037)、放线菌科(OR=1.561,95%CI:1.027-2.374,P=0.037)、放线菌属(OR=1.544,95%CI:1.130-2.110,P=0.006)、丁酸球菌属(OR=1.781,95%CI:1.194-2.657,P=0.005)、粪球菌属-2(OR=1.550,95%CI:1.068-2.251,P=0.021)、Family ⅩⅢ UCG-001属(OR=1.473,95%CI:1.001-2.168,P=0.049)、产甲烷短杆菌属(OR=1.274,95%CI:1.001-1.621,P=0.049)、罗氏菌属(OR=1.429,95%CI:1.015-2.013,P=0.041)的丰度升高,会增加患者骨质疏松性骨折的风险;②拟杆菌纲(OR=0.660,95%CI:0.455-0.959,P=0.029)、拟杆菌目(OR=0.660,95%CI:0.455-0.959,P=0.029)、克里斯滕森氏菌科(OR=0.725,95%CI:0.529-0.995,P=0.047)、瘤胃球菌科(OR=0.643,95%CI:0.443-0.933,P=0.020)、肠杆菌属(OR=0.558,95%CI:0.395-0.788,P=0.001)、直肠真杆菌属(OR=0.631,95%CI:0.435-0.916,P=0.016)、毛螺菌科-UCG008(OR=0.738,95%CI:0.546-0.998,P=0.048)、瘤胃梭菌属-9(OR=0.492,95%CI:0.324-0.746,P=0.001)的丰度升高,会降低患者骨质疏松性骨折的风险。③文章通过孟德尔随机化方法发现了16种与骨质疏松性骨折相关的肠道菌群,即以肠道菌群为暴露因素,骨质疏松性骨折为结局变量,8种肠道菌群与骨质疏松性骨折呈正向因果关联,另外8种肠道菌群与骨质疏松性骨折呈负向因果关联。④此研究结果不仅为临床上骨质疏松性骨折的早期预测及潜在治疗靶点确定了新的生物标志物,还为骨组织工程中研究通过肠道菌群改善骨质疏松性骨折的发生与预后提供了实验基础和理论依据。展开更多
背景:既往研究表明,中老年人体内组织蛋白酶K水平可通过影响骨密度来干预骨质疏松的发生和发展,但组织蛋白酶家族与其他人群骨密度之间是否存在因果关系仍未知。目的:探讨组织蛋白酶与骨密度的因果关系。方法:从IEU Open GWAS数据库提取...背景:既往研究表明,中老年人体内组织蛋白酶K水平可通过影响骨密度来干预骨质疏松的发生和发展,但组织蛋白酶家族与其他人群骨密度之间是否存在因果关系仍未知。目的:探讨组织蛋白酶与骨密度的因果关系。方法:从IEU Open GWAS数据库提取与8种组织蛋白酶相关的遗传位点作为工具变量,以5个年龄段人群的骨密度作为结局。通过双向孟德尔随机化分析,评估组织蛋白酶与骨密度的因果关系。使用Cochran’s Q检验评估遗传工具变量的异质性,使用MR-Egger截距检验评估多效性,使用留一法评估作为工具变量的单核苷酸多态性对暴露和结局因果关系影响的敏感性。结果与结论:①正向孟德尔随机化的逆方差加权法结果显示,组织蛋白酶H与>45岁且≤60岁人群的骨密度呈负相关[OR(95%CI)=0.965(0.94-0.99),P=0.04],组织蛋白酶Z与>30岁且≤45岁人群的骨密度呈负相关[OR(95%CI)=1.06(1.00-1.11),P=0.03];②敏感性分析结果显示因果关系稳定,MR-Egger截距分析未检测到潜在的水平多效性;③反向孟德尔随机化结果显示,骨密度对组织蛋白酶无显著反向作用;④上述结果证实,组织蛋白酶对部分年龄段人群骨密度会造成影响,可能会增加骨质疏松症的发病风险,应给予更多关注。展开更多
The diagnosis of herbal hepatotoxicity or herb induced liver injury(HILI) represents a particular clinical and regulatory challenge with major pitfalls for the causality evaluation.At the day HILI is suspected in a pa...The diagnosis of herbal hepatotoxicity or herb induced liver injury(HILI) represents a particular clinical and regulatory challenge with major pitfalls for the causality evaluation.At the day HILI is suspected in a patient,physicians should start assessing the quality of the used herbal product,optimizing the clinical data for completeness,and applying the Council for International Organizations of Medical Sciences(CIOMS) scale for initial causality assessment.This scale is structured,quantitative,liver specific,and validated for hepatotoxicity cases.Its items provide individual scores,which together yield causality levels of highly probable,probable,possible,unlikely,and excluded.After completion by additional information including raw data,this scale with all items should be reported to regulatory agencies and manufacturers for further evaluation.The CIOMS scale is preferred as tool for assessing causality in hepatotoxicity cases,compared to numerous other causality assessment methods,which are inferior on various grounds.Among these disputed methods are the Maria and Victorino scale,an insufficiently qualified,shortened version of the CIOMS scale,as well as various liver unspecific methods such as thead hoc causality approach,the Naranjo scale,the World Health Organization(WHO) method,and the Karch and Lasagna method.An expert panel is required for the Drug Induced Liver Injury Network method,the WHO method,and other approaches based on expert opinion,which provide retrospective analyses with a long delay and thereby prevent a timely assessment of the illness in question by the physician.In conclusion,HILI causality assessment is challenging and is best achieved by the liver specific CIOMS scale,avoiding pitfalls commonly observed with other approaches.展开更多
基金Project supported by the Key National Natural Science Foundation of China(Grant No.62136005)the National Natural Science Foundation of China(Grant Nos.61922087,61906201,and 62006238)。
文摘Physics-informed neural networks(PINNs)have become an attractive machine learning framework for obtaining solutions to partial differential equations(PDEs).PINNs embed initial,boundary,and PDE constraints into the loss function.The performance of PINNs is generally affected by both training and sampling.Specifically,training methods focus on how to overcome the training difficulties caused by the special PDE residual loss of PINNs,and sampling methods are concerned with the location and distribution of the sampling points upon which evaluations of PDE residual loss are accomplished.However,a common problem among these original PINNs is that they omit special temporal information utilization during the training or sampling stages when dealing with an important PDE category,namely,time-dependent PDEs,where temporal information plays a key role in the algorithms used.There is one method,called Causal PINN,that considers temporal causality at the training level but not special temporal utilization at the sampling level.Incorporating temporal knowledge into sampling remains to be studied.To fill this gap,we propose a novel temporal causality-based adaptive sampling method that dynamically determines the sampling ratio according to both PDE residual and temporal causality.By designing a sampling ratio determined by both residual loss and temporal causality to control the number and location of sampled points in each temporal sub-domain,we provide a practical solution by incorporating temporal information into sampling.Numerical experiments of several nonlinear time-dependent PDEs,including the Cahn–Hilliard,Korteweg–de Vries,Allen–Cahn and wave equations,show that our proposed sampling method can improve the performance.We demonstrate that using such a relatively simple sampling method can improve prediction performance by up to two orders of magnitude compared with the results from other methods,especially when points are limited.
文摘An improved safety analysis based on the causality diagram for the complex system of micro aero-engines is presented.The study is examined by using the causality diagram in analytical failure cases due to rupture or pentration in the receiver of micro turbojet engine casing,and the comparisons are also made with the results from the traditional fault tree analysis.Experimental results show two main advantages：（1）Quantitative analysis which is more reliable for the failure analysis in jet engines can be produced by the causality diagram analysis;（2）Graphical representation of causality diagram is easier to apply in real test cases and more effective for the safety assessment.
文摘背景:膝骨关节炎是一种常见的关节软骨及周围组织损伤的慢性炎症性疾病,而免疫细胞在膝骨关节炎免疫炎症反应中起到重要作用,但其中的具体机制仍有待深入研究。目的:采用孟德尔随机化方法来评估731种免疫细胞表型与膝骨关节炎风险之间的潜在因果关系。方法:使用全基因组关联分析(GWAS)目录中公开获取731种免疫细胞表型的全基因组关联分析统计数据(从GCST0001391到GCST0002121)和IEUGWAS数据库中膝骨关节炎的全基因组关联分析数据(ebi-a-GCST007090)。采用逆方差加权法、MR-Egger回归法、加权中位数法、加权模型法和简单模型法来研究免疫细胞与膝骨关节炎之间的因果关系。敏感性分析用于检验孟德尔随机化分析结果是否可靠,然后以同样方法进行反向孟德尔随机化分析。结果与结论:①正向分析结果表明,共有4种免疫细胞表型与膝骨关节炎有显著的因果关系(FDR<0.20),其中B细胞中的CD27 on CD24+CD27+(OR=1.026,P=0.00026,Pfdr=0.18)、髓系细胞中的CD33 on CD33dim HLA DR-(OR=1.014,P=0.00050,Pfdr=0.18)以及Treg细胞中的CD45RA+CD28-CD8br%CD8br(OR=1.001,P=0.00078,Pfdr=0.18)与膝骨关节炎风险呈直接的正向因果关联;单核细胞中PDL-1 on monocyte(OR=0.952,P=0.00098,Pfdr=0.18)与膝骨关节炎风险呈直接的负向因果关联。②反向分析结果表明,当膝骨关节炎作为暴露数据时,与731种免疫细胞表型均不具有显著因果关系(FDR<0.20)。③敏感性分析结果显示:双向孟德尔随机化的Cochran’s Q检验和MR-Egger回归法结果P值均大于0.05,表明免疫细胞表型与膝骨关节炎之间的因果效应分析不存在显著的异质性和多效性。④上述结果证实,CD27 on CD24+CD27+,CD33 on CD33dim HLA DR-,CD45RA+CD28-CD8br%CD8br以及PDL-1 on monocyte免疫细胞表型与膝骨关节炎之间可能具有较为显著的潜在因果关系,这为研究膝骨关节炎的生物学机制及探索膝骨关节炎的早期防治提供有价值的线索,也为干预性药物的开发提供了新的方向。
文摘背景:骨质疏松性骨折是骨质疏松症最严重的并发症,既往的研究已经证实了肠道菌群对骨骼组织具有调节作用,肠道菌群与骨质疏松性骨折有着重要关系,但是二者之间的因果关系尚不清楚。目的:使用孟德尔随机化(MR)方法探索肠道菌群与骨质疏松性骨折之间的因果关系。方法:从IEU Open GWAS数据库和芬兰数据库R9中分别获得了肠道菌群和骨质疏松性骨折的GWAS数据集,以肠道菌群作为暴露因素,骨质疏松性骨折作为结局变量,采用随机效应逆方差加权法、MR-Egger回归、加权中位数法、简单模型法以及加权模型法进行孟德尔随机化分析来评估肠道菌群与骨质疏松性骨折之间是否存在因果关系,通过敏感性分析来检验结果的可靠性和稳健性,并进行反向孟德尔随机化分析来进一步验证正向孟德尔随机化分析中确定的因果关系。结果与结论:①此孟德尔随机化分析结果表明,肠道菌群与骨质疏松性骨折之间存在因果关系。放线菌目(OR=1.562,95%CI:1.027-2.375,P=0.037)、放线菌科(OR=1.561,95%CI:1.027-2.374,P=0.037)、放线菌属(OR=1.544,95%CI:1.130-2.110,P=0.006)、丁酸球菌属(OR=1.781,95%CI:1.194-2.657,P=0.005)、粪球菌属-2(OR=1.550,95%CI:1.068-2.251,P=0.021)、Family ⅩⅢ UCG-001属(OR=1.473,95%CI:1.001-2.168,P=0.049)、产甲烷短杆菌属(OR=1.274,95%CI:1.001-1.621,P=0.049)、罗氏菌属(OR=1.429,95%CI:1.015-2.013,P=0.041)的丰度升高,会增加患者骨质疏松性骨折的风险;②拟杆菌纲(OR=0.660,95%CI:0.455-0.959,P=0.029)、拟杆菌目(OR=0.660,95%CI:0.455-0.959,P=0.029)、克里斯滕森氏菌科(OR=0.725,95%CI:0.529-0.995,P=0.047)、瘤胃球菌科(OR=0.643,95%CI:0.443-0.933,P=0.020)、肠杆菌属(OR=0.558,95%CI:0.395-0.788,P=0.001)、直肠真杆菌属(OR=0.631,95%CI:0.435-0.916,P=0.016)、毛螺菌科-UCG008(OR=0.738,95%CI:0.546-0.998,P=0.048)、瘤胃梭菌属-9(OR=0.492,95%CI:0.324-0.746,P=0.001)的丰度升高,会降低患者骨质疏松性骨折的风险。③文章通过孟德尔随机化方法发现了16种与骨质疏松性骨折相关的肠道菌群,即以肠道菌群为暴露因素,骨质疏松性骨折为结局变量,8种肠道菌群与骨质疏松性骨折呈正向因果关联,另外8种肠道菌群与骨质疏松性骨折呈负向因果关联。④此研究结果不仅为临床上骨质疏松性骨折的早期预测及潜在治疗靶点确定了新的生物标志物,还为骨组织工程中研究通过肠道菌群改善骨质疏松性骨折的发生与预后提供了实验基础和理论依据。
文摘背景:既往研究表明,中老年人体内组织蛋白酶K水平可通过影响骨密度来干预骨质疏松的发生和发展,但组织蛋白酶家族与其他人群骨密度之间是否存在因果关系仍未知。目的:探讨组织蛋白酶与骨密度的因果关系。方法:从IEU Open GWAS数据库提取与8种组织蛋白酶相关的遗传位点作为工具变量,以5个年龄段人群的骨密度作为结局。通过双向孟德尔随机化分析,评估组织蛋白酶与骨密度的因果关系。使用Cochran’s Q检验评估遗传工具变量的异质性,使用MR-Egger截距检验评估多效性,使用留一法评估作为工具变量的单核苷酸多态性对暴露和结局因果关系影响的敏感性。结果与结论:①正向孟德尔随机化的逆方差加权法结果显示,组织蛋白酶H与>45岁且≤60岁人群的骨密度呈负相关[OR(95%CI)=0.965(0.94-0.99),P=0.04],组织蛋白酶Z与>30岁且≤45岁人群的骨密度呈负相关[OR(95%CI)=1.06(1.00-1.11),P=0.03];②敏感性分析结果显示因果关系稳定,MR-Egger截距分析未检测到潜在的水平多效性;③反向孟德尔随机化结果显示,骨密度对组织蛋白酶无显著反向作用;④上述结果证实,组织蛋白酶对部分年龄段人群骨密度会造成影响,可能会增加骨质疏松症的发病风险,应给予更多关注。
文摘The diagnosis of herbal hepatotoxicity or herb induced liver injury(HILI) represents a particular clinical and regulatory challenge with major pitfalls for the causality evaluation.At the day HILI is suspected in a patient,physicians should start assessing the quality of the used herbal product,optimizing the clinical data for completeness,and applying the Council for International Organizations of Medical Sciences(CIOMS) scale for initial causality assessment.This scale is structured,quantitative,liver specific,and validated for hepatotoxicity cases.Its items provide individual scores,which together yield causality levels of highly probable,probable,possible,unlikely,and excluded.After completion by additional information including raw data,this scale with all items should be reported to regulatory agencies and manufacturers for further evaluation.The CIOMS scale is preferred as tool for assessing causality in hepatotoxicity cases,compared to numerous other causality assessment methods,which are inferior on various grounds.Among these disputed methods are the Maria and Victorino scale,an insufficiently qualified,shortened version of the CIOMS scale,as well as various liver unspecific methods such as thead hoc causality approach,the Naranjo scale,the World Health Organization(WHO) method,and the Karch and Lasagna method.An expert panel is required for the Drug Induced Liver Injury Network method,the WHO method,and other approaches based on expert opinion,which provide retrospective analyses with a long delay and thereby prevent a timely assessment of the illness in question by the physician.In conclusion,HILI causality assessment is challenging and is best achieved by the liver specific CIOMS scale,avoiding pitfalls commonly observed with other approaches.
