Diabetes is becoming one of the most frequent and serious diseases in INDIA and across the world. It is not only damaging to the blood, but it also causes a variety of disorders such as blindness, renal disease, kidney problems, heart disease, and other conditions that result in a large number of deaths each year. As a result, it is critical to design a system that can accurately detect diabetes patients using medical information. By training its properties in a five-fold and ten-fold cross-validation method, Using deep neural networks, we propose an approach for diabetes diagnosis. The data set for Pima Indian Diabetes (PID) was obtained from the unique client identifier (UCI) machine learning depository database. With a prediction accuracy of 98.35 percent, Fivefold cross-validation had an F1 result of 98 and an Medical Counseling Committee (MCC)of 97, the outcome on the Pelvic inflammatory disease (PID) dataset show that Deep Learning can construct an advantageous system for diabetes prediction. Furthermore, 97.11% accuracy and sensitivity Ten- fold cross-validation yielded a sensitivity of 96.25% and a specificity of 98.80%. The results of the experiments reveal that the suggested strategy gives promising results when employing five-fold cross-validation.

Artificial Neural Network, Binary Space SeparatingDiabetes, Hierarchical Neuro-Fuzzy , Matthews Correlation Coefficient, Receiver Operating Characteristi