Affiliations: [a] Medical Informatics Laboratory, Democritus University of Thrace, 68100, Greece | [b] Lab of Forest Informatics, Democritus University of Thrace, Pandazidou 193 str, Orestias, PC 68200, Greece | University of Patras, School of Engineering, Dept of Computer Engineering & Informatics, 26500 Patras, Greece
Abstract: This study examines a hybrid Artificial Intelligence modelling approach in terms of its classification efficiency in the abdominal pain disease, especially in childhood. The classification model consists of a series of efficient Artificial Neural Network (ANN) architectures which have been evaluated by the application of an innovative Fuzzy Algebraic Information System (FAIS) [10]. FAIS offers a flexible approach by employing fuzzy sets and relations, fuzzy intensification and dilution techniques towards the assessment of neural models under different levels of accuracy and under various perspectives. The fact that FAIS produces an overall ANN evaluation index and also individual partial evaluation indices corresponding to each separate output neuron, makes it very useful for the specific disease where even the slightest error can cause the unnecessary operative treatment of the disease. In the examined cases, the produced ANN models have proven to perform classification with success. The whole approach comprises of a mixture of ANN development techniques together with Fuzzy modelling functions and relations and thus it can be considered as a Hybrid one.
