Previous Models to Identify Predictors
Other investigators have attempted to identify predictors of recurrent
MR. Some studies have been limited by performing only univariate
analysis to identify specific echocardiographic measurements and
clinical features associated with recurrent MR (Table). Others, using
traditional statistical modelling, have been able to predict recurrent
MR with AUC values of 0.82, sensitivity of 0.86 and specificity of 0.70.
[3] but were limited to the inclusion of only 10 clinical and
echocardiographic variables (Table). Though the above modelling of
recurrent MR have been successful in provided us with mechanistic and
clinical insights, they have not been reproducible and sometimes require
extremely complicated measurements that are not routinely performed in
echocardiography or collected in clinical practice. As such, a tool
built on data that is easily and ubiquitously collected and available
from all patients becomes important to inform all surgeons on the best
surgical plan.