The Sensitivity Analysis For Longitudinal Missing Data And Corresponding Effects Of Intervention In Cardiac Rehabilitation

In a longitudinal study of comprehensive cardiac rehabilitation for secondary prevention intervention model, owing to the reasons of the variable of monitoring data is numerous and collection process is more complex,uncooperative subjects, the physically challenged and so all, inevitably makes missing data. If we analysis the missing data directly using the methods for complete data or filling it without considering missing mechanism and strongly influence subject,we are all tends to be a biased estimates, even wrong conclusions. To more fully utilizing and revealing the hidden information of clinical monitoring missing data,we mainly introduce the DK selection models and the method of sensitivity analysis applied in identified missing mechanism, how to use it to identify missing mechanism and finding influence subject for different missing data.We simulate the data with different sample sizes and different proportions of missing and clarify the approach how to use DK selection model identify missing mechanism combined the data of coronary artery syndrome comprehensive intervention for secondary prevention of cardiac rehabilitation and introduce the way of recognizing influence subject of three missing mechanism. The whole process is implemented by SAS software programming. After simulation studies we confirm that identified different mechanisms needs various sample size and different proportion missing data, DK select models can accurately identify MCAR and MAR, but it is not suitable to identify the MNAR mechanism. When we intend to recognize the MNAR, we should remove some influence subject and then conduct sensitivity analysis. In this paper, we collect133acute coronary syndrome case data about SCL-90, SAS who received secondary comprehensive intervention for cardiac rehabilitation and come from a cardiovascular specialist hospital CCU ward. We further validate using DK selection model combined local influence tool based removing the influence subject can identify the missing mechanism. Finally, we get the missing mechanism is MCAR. After filling data using MCMC strategy, we can confirm that ignoring the mechanisms tends to be a biased estimates.In summary, this paper clarify how to use DK selection model and sensitivity analysis identified the missing mechanism for the problem of missing data in longitudinal studies.We confirm different sample size and various missing proportion all have impact on identified missing mechanism through using simulation research methods. We use local influence tool removing strong influence subject for sensitivity analysis through SAS software programming and collect the longitudinal data of CCU cardiac secondary prevention interventions rehabilitation patients with acute coronary syndrome.Finally,we identify missing mechanism finding strong influence subject and analysis longitudinal studies missing data.