Closed KateJohnson closed 5 years ago
Get ORs of diagnosis for case detection methods
For every patient in CanCold: p * x = y p = predicted probability from logistic regression x= OR for the effect of case detection on diagnosis mean(y)= sensitivity of case detection method
The sensitivity and specificity of 4 case detection methods are defined in EPIC:
input$values$diagnosis$case_detection_methods
None CDQ195 CDQ165 FlowMeter FlowMeter_CDQ
[1,] 0 0.4742 0.6069 0.5676 0.5366
[2,] 0 0.1858 0.4115 0.0735 0.0134
To change the case detection method during the model run, change the inputs as follows:
input <- init_input()
input$values$diagnosis$logit_p_diagnosis_by_sex <- cbind(male=c(intercept=-5, age=-0.0324, smoking=0.3711, fev1=-0.8032, gpvisits=0.0087, cough=0.208, phlegm=0.4088, wheeze=0.0321, dyspnea=0.722, case_detection=input$values$diagnosis$case_detection_methods[1,"CDQ195"]), female=c(intercept=-5-0.4873, age=-0.0324, smoking=0.3711, fev1=-0.8032, gpvisits=0.0087, cough=0.208, phlegm=0.4088, wheeze=0.0321, dyspnea=0.722, case_detection=input$values$diagnosis$case_detection_methods[1,"CDQ195"]))
Only a placeholder for now. The OR should be from the effectiveness of a case detection method (can be from SN and SP).