MARおよびMNAR下での標的学習の枠組み内での縦断的ランダム化臨床試験における欠損値の処理

Missing data is a common issue in longitudinal randomized clinical trials (RCTs), where the effect of treatment is estimated by a pre-specified statistical model. The targeted learning approach, through targeted maximum likelihood estimation (TMLE), has been suggested and evaluated in both clinical trials and observational studies to deal with missing data under the assumption of missing at random (MAR). In this research project, we propose methods for handling missing data in longitudinal RCTs Within the framework of targeted learning through longitudinal maximum likelihood estimation (LTMLE), with missing data mechanisms MAR and missing not at random (MNAR). The methods are applied to a real public RCT dataset.