Perinatal asphyxia: Development of an ‘Independence Bayes’ computer prediction model

Summary

The development of an ‘Independence Bayes’ model for the prediction of birth asphyxia from both clinical and non-clinical variables is described, including the necessary steps in validation of the original clinical database.

Predictions using the model were stable when conditional probabilities were derived from as few as 2000 cases. Some variation in predictive performance was noted with time; the probable reasons for this are discussed.