Adaptive bio-inspired gene optimisation based deep neural associative classification for diabetic disease diagnosis

Abstract

Associative classification plays a significant role in data mining. With several classification techniques being used, the accuracy with which classification was performed was found to be inadequate. To overcome this issue, an adaptive bio-inspired gene optimisation based deep neural associative classification (ABGO-DNAC) technique is proposed. ABGO-DNAC technique generates association rules with the minimal number of medical attributes by applying ABGO algorithm and choosing optimal attributes from the medical dataset. With formulated association rules, Gaussian deep feed forward neural learning (GDFNL) is designed for diabetic disease classification. GDFNL deeply analyses the patient's medical data and classify patients as normal or abnormal. Simulation evaluation of ABGO-DNAC technique is performed on disease prediction accuracy, disease prediction time and false positive rate with different patients. Simulation results depict ABGO-DNAC technique disease prediction accuracy and also reduce diabetic disease diagnosing as compared to state-of-the-art works.