Replication of an association study and validation of three predictive models of proliferative vitreo-retinopathy

Following the association study and based on its genotype data, we identified
risk genetic markers of PVR and set up three predictive models. Machine-
learning methods have been used to predict the probability of developing PVR
after primary rhegmatogenous retinal detachment (RD) using subsets of 197 SNPs
in 30 candidate genes. The model is developed so that it optimally fits the
data and predicts the patients* outcome in the data set as accurately as
possible, using the original data set. Previous to applying predictive models
in clinical practices, it is necessary to establish if the model offers
realistic estimations. The generalizability of models or external validation
will be proved by a new sample, the validation sample.

To perform an external (geographic and temporal) validation of predictive
models for estimating the probability of PVR following primary RD using a
subgroup of SNPs from a total of 197 in 30 candidate genes previously studied.

Case control validation study.

Participants do not benefit. Risks are negligible.