Diabetes Diagnosis Using Fuzzy – Neuro Hybrid Control Model
Danladi Ali *
Department of Pure and Applied Physics, Adamawa State University, Mubi, Nigeria
*Author to whom correspondence should be addressed.
Abstract
Diabetes is caused due to an inability of a body to produce or respond to hormone insulin causing abnormal metabolism of carbohydrate which can lead to rising in sugar level in the blood. This work proposed a fuzzy - neuro hybrid control model to diagnose diabetes in terms of seven symptoms such as an increase in urination, increase in thirst, increase in fatigue, tingling in hands/feet, blurred vision, sores slow to heal and significant loss of weight. 15 patients were diagnosed with sugar levels as followed 9.6 mmol/l, 6.8 mmol/l, 9.1 mmol/l, 11.2 mmol/l, 6.5 mmol/l, 5.7 mmol/l, 11.8mmol/l, 8.9 mmol/l, 7.0 mmol/l, 11.0 mmol/l, 8.5 mmol/l, 9.0mmol/l, 12.4 mmol/l, 9.5 mmol/l and 10.4 mmol/l. The average diagnosis error is obtained as 0.05%, which is acceptable in medical diagnosis. In this regards, it is recommended that fuzzy- neuro hybrid control model is a good soft computing tool for diagnosing diabetes.
Keywords: Diabetes, soft computing, fuzzy logic, neural network, sugar level, expert domain