Expert System for Diagnosis Fever with Rash Diseases for Pediatric Age Group using Adaptive Neuro- Fuzzy Inference System

Fever accompanied by a skin rash is the most common disease of the pediatric age group. They are diverse and characterized by similar symptom, which makes it difficult to distinguish between them. This work presents an Adaptive Neuro-Fuzzy Inference System (ANFIS) to diagnoses five diseases (measles, rubella, infantile rosella, scarlet fever, and chickenpox). Dataset about symptoms and diagnosis were collected from specialized physicians then they were analyzed using the Chi-Square test to identify symptoms that play a significant role in the diagnosis process including shared symptoms like temperature, type of rash and its location. In addition, some several distinctive signs such as Koplik's spots, lymphadenopathy, tongue color and itching. This paper proposed a hybrid method between Mamdani's fuzzy inference controller and multi-layered neural network to overcome the drawbacks of using each technique alone and enhance the classification process. Finally, the model was validated using 250 real-world cases and the experimental results showed that the proposed system succeeded in detecting and recognizing diseases with high accuracy close to that of human diagnosis with an average accuracy of 95.50%. The results were compared with the accuracy obtained using the fuzzy logic technique separately and with the results of previous studies.