An Optimized Deep Learning-based framework for predicting diabetes mellitus using FFNN

Diabetes mellitus (DM) is a major public health problem in Egypt, and the illness is regarded as a contemporary epidemic across the world. Diabetes is becoming more common, which is a cause for serious concern. As a result, precise and timely identification of the illness is critical. Health and research institutions have also recently expressed a serious interest in developing and implementing cutting-edge healthcare systems. Therefore, it is necessary to accurately and quickly identify the condition. To solve this issue, scientific research has been carried out, but the outcomes have fallen short. Four layers make up the proposed Diabetes mellitus prediction with deep learning (DMPDL) framework structure. The DMPDL framework proposes the Gaussian Modification of Grey Wolf Optimization Algorithm (GMGWO), which is used to identify the ideal subset of characteristics and reduce classification error. It is crucial to offer an accurate optimizer that can be used to predict diabetic disorders more quickly and accurately if this framework is provided to address such concerns. The GMGWO was created to identify new areas to look for. Following that, it is examined using the CEC2017 benchmark features. The experimental findings show that the DMPDL framework outperforms the others in all circumstances, with an accuracy of 82.72% for the Pima Indian Diabetes (PIDM) dataset, and 97.63% for the Early-Stage Diabetes Risk (ESDR) dataset