Selecting The Optimal Treatment Using CSTE Curve Under Longitudinal Data

Treatment selection based on patient's biomarker has been widely concerned in biomedical research.Based on the independent data,many scholars proposed methods including the use of SIC(select effect curve),CSTE(covariate-specific treatment effect)curve to select the best treatment.Considering that longitudinal data is a kind of important data type,it has been widely used in biomedical field in recent years,and there are many clinical trials which record the experimenter repeatedly.If you give up this part of information,it will cause great loss.Therefore,this paper studies how to use CSTE curve to select the optimal treatment under longitudinal data.In this paper,the generalized estimation equation method based on spline is used to estimate the CSTE curve,and the asymptotic property of CSTE curve estimation is proved.Based on the asymptotic normality of the CSTE curve,we also give the estimation of the confidence band of the CSTE curve.The results of numerical simulation and case analysis show that the CSTE curve and its confidence band are effective...