w******8
发帖数: 59 | 1 Here we have N subjects, each subject having a few times of measurement of a
predictor X and a outcome variable Y (binary).
If each subject were measured only once I could have done a ROC curve to see
if X has any prediction value for Y, but since now each subject is having
multiple measurements (from different time point), is it possible to do a
ROC analysis?
Thanks! | P****D
发帖数: 11146 | | w******8
发帖数: 59 | 3 Here we have N subjects, each subject having a few times of measurement of a
predictor X and a outcome variable Y (binary).
If each subject were measured only once I could have done a ROC curve to see
if X has any prediction value for Y, but since now each subject is having
multiple measurements (from different time point), is it possible to do a
ROC analysis?
Thanks! | P****D
发帖数: 11146 | | w******4
发帖数: 488 | 5 比较简单一些的做法是对每一个subject根据repeated measure算出一个值（指标），
可以是mean, median, iqr etc., 然后根据这个值去预测y， 画出ROC |
|
