An integrated hybrid data mining system for preterm birth risk assessment based on a 'Semantic Web services for healthcare' framework

A re-engineered approach to the early prediction of preterm birth is presented as a complementary technique to the current procedure of using costly and invasive clinical testing on high-risk maternal populations only. The preterm birth risk assessment tool is based on an integrated hybrid data mining screening system developed using a combination of: Artificial Neural Networks (ANNs), specifically a classification-based Multilayer Perceptron (MLP) ANN and a MLP ANN with risk stratification; case based reasoning; and a decision tree voting algorithm. In developing the integrated hybrid data mining system, the classification-based MLP ANN was employed as an outcome prediction tool for estimating preterm birth and high-risk preterm birth on a heterogeneous maternal population; risk estimations use obstetrical variables available to physicians prior to 23 weeks gestation. The trained hybrid system yields a sensitivity of 65% and a specificity of 84%, meeting the clinical criteria determined by collaborating physicians; the system is successfully validated on 9701 new patient cases.; The developed tool is part of a larger goal: to offer perinatal risk assessment to physicians, researchers, and patients via a 'Semantic Web service for healthcare' framework. The framework is based on Semantic Web technology; a diagnostic clinical decision support ontology for perinatal care is created, attaching meaning to all resources. Candidate Web services are defined and service interactions are specified through composition templates, the service integration architecture is proposed.