诊断试验(英文)课件

,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,*,单击此处编辑母版标题样式,Analysis and evaluation of the evidence of diagnostic test,Clinical Trail Study Center,Cao,Sumei,Diagnostic test are not just about diagnosis,Screening,Determining severity,Optimally therapy,Prognosis,Monitor,Example,Carotid ultrasound can tell you the,severity,of the patients carotid stenosis,Carotid ultrasound can tell you the patients,prognosis,for stroke and death,Carotid ultrasound can predict your patients likely,responsiveness to therapy,Basic principles of conducting diagnostic studies,Apply the gold standard to determine whether or not the target condition is present,Gold standard:The most recognized standard for clinician to diagnose the target condition,Pathological measurement,Operation finding,Special imaging detection,Long-term follow-up,Recognized standard,What if your test is more gold than the standard,May lead to underestimate of the diagnostic power of the evaluating,One strategy for dealing with this problem is to use long-term follow-up as a gold standard,To Whom Should the Gold Standard Be Applied?,to everyone,selective performing the gold standard on patients may result in“verification bias”or“workup bias”,Recruit your participants,Recruit the target-negative and target-positive participants identified by gold standard,characteristic of those to whom you will want to apply the test in clinical practice,Including a broad spectrum of the diseased,case,：,from mildly to severely,control,：,a broad spectrum of competing conditions,An alternative approach is that,recruit,ing,a consecutive sample of patients,Measurement procedures,Specifying test technique,Reproducibility,Blinding of the individual conducting or interpreting the test to the gold standard,Select statistical procedure,Calculating sample size,Example:Assuming a sensitivity of 80%,specificity of 60%of ultrasonography for diagnosis of cholecystolithiasis.How many samples are needed?,Result evaluation index,Example:126 patients underwent independent,blind BNP measurement and echocardiography for diagnosis of LVD.,sensitivity,：,a/(a+c)=35/40=0.88,or 88%,specificity,：,d/(b+d)=29/86=0.34,or 34%,positive predictive value(PPV),：,a/(a+b)=35/92=0.38,or 38%,negative predictive value(NPV),：,d/(c+d)=29/34=0.85,or 85%,prevalence:(a+c)/(a+b+c+d)=40/126=0.32,or 32%,Pre-test odds:pre-test probability/(1-pre-test probability)=32%/68%=0.47,positive likelihood ratio(LR+):Sen/1-Spe=88%/(100%-34%)=1.3,Multilevel likelihood ratios,Stability of the index,Stable:Sen,Spe,Relatively stable:LR+,LR-,Unstable:PPV,NPV,prevalence:,Receiver operating characteristic,curves(ROC,),It illustrates the performance of a diagnostic test when you select different cut-points to distinguish“normal”from“abnormal”,It demonstrates the fact that any increase in sensitivity will be accompanied by a decrease in specificity,The closer the curve gets to the upper left corner of the display,the more the overall accuracy of the test,The closer the curve comes to the 45-degree diagonal of the ROC space,the less accurate the test,The area under the curve provides an overall measure of a tests accuracy,Fig A ROC for BNP as a diagnostic test for LVD,Parallel test,A,test B test Result,+,+,+,+,-,Reduction miss diagnosis,Exclude some disease,When prevalence is low,as the primary screening method,Serial test,A,test B test Result,+,+,-,+-,-,Sen,=,Sen,A,SenB,Spe,=,Spe,A+(1-Spe A),Spe,B,Misdiagnosis may cause nuisance effect,Confirmatory diagnosis,Serial test with enzyme labeled compound assay for diagnosis of myocardial infarction,Enzyme labeled compound assay,Sen,Spe,CPK,96,57,SGOT,91,74,LDH,87,91,91,Multivariate analysis,SEN,SPE,single variable analysis,marker,methods,SEN,(%),spe,(%),cutoff,AREA,a,ELISA,90.7,88.8,0.1915,0.926,b,ELISA,77.3,73.2,0.2035,0.802,c,ELISA,74.2,70.9,0.0905,0.762,d,ELISA,78.4,81.6,1.08,0.836,e,ELISA,90.7,84.4,0.356,0.932,f,ELISA,84.5,81.6,0.799,0.899,multivariate analysis using logistic regression,Combined markers,SEN,（）,SPE,（）,AREA,a and,b,88.8,82.5,0.926,a and c,87.7,82.5,0.927,a and e,91.6,90.07,0.974,a and f,95.5,90.07,0.967,a and d,87.2,85.6,.0936,b,and c,78.8,76.3,0.837,b,and d,87.7,86.6,0.934,b,and e,83.8,82.5,0.900,b,and f,82.7,81.4,0.863,C and d,87.7,85.6,0.946,C and e,88.3,85.6,0.926,C and f,81.6,79.4,0.854,d and e,89.4,86.6,0.946,d and f,88.3,86.6,0.952,e and f,87.7,85.6,0.933,Prediction the probability of a disease,Logit(P,)=-0.934+4.797,x,a,+2.203,x,e,Avoiding,overfitting,Overfitting,occurs when a computer model identifies a“chance”pattern that discriminates cancer patients from non-cancer patients,perfectly fitting that dataset but not reproducible in other data sets.,One way to avoiding,overfitting,is to randomly split the data into separate training and test samples.,The EBM steps for diagnostic tests,Looking for the most suitable study papers according to the clinical question,Bring forward the question in clinic,Example 2:,if,detection,of,serum,forritin,can,diagnose,Iron deficiency anemia?,Search the c