Classification and regression tree analysis for predicting visual outcome after open-globe injuries in Siriraj Hospital
Objective:
To create a model for predicting visual outcome after open-globe injuries by using data of Siriraj Hospital.
Material And Method:
Retrospective data of patients presented with open-globe injuries between January 2007 and December 2010 were used to create prognostic model. Seventeen factors at initialpresentation were collected and evaluated to develop the model by mean of Classification and Regression Tree analysis (CART). The prognostic tree was validated by using the sample ofopen-globe patients who presented between January 2011 and July 2011.
Results:
The information of 231 eyes from 230 patients was analyzed to create a classification tree model. The calculated model composed of the two greatestpredictivefactors, no light perception (NPL), andpresence of relative afferentpupillary defect (RAPD). No patient with NPL at initial examination had vision at the six-month follow-up period. The other patients could be classified and predicted vision by using the presence of RAPD.
Conclusion:
The classification tree model developed in the present study is easy to calculate and has major significant predictive outcome for the open-globe injured patients.