An Approach using Certainty Factor Rules for Aphasia Diagnosis

Artificial Intelligence methods are frequently applied to medical domains assisting in various tasks, like diagnosis. The implementation of corresponding intelligent systems is based on available datasets and expert knowledge. In this paper, we present a rule-based approach used for aphasia diagnosis. The approach uses certainty factor rules created from a dataset of records involving persons diagnosed with aphasia. The employed certainty factor formalism is an extension to a previous certainty factor formalism. To model each case, nineteen attributes are used. Seven of them are discrete and the rest of them are integer. Experiments were performed testing the performance of the specific rule-based approach, a decision tree method and a feedforward neural network. Experimental results show that the rule-based approach performs well and requires only three of the eighteen input attributes to produce an output. To the best of our knowledge, there is no other published approach using certainty factor rules for aphasia diagnosis.