This study introduces a novel deep learning (DL) technique for early diabetes prediction with high accuracy. The model, based on a multi-layer perceptron (MLP) algorithm within a deep neural network (DNN), consists of ten hidden layers and multiple epochs. The hyperparameters were fine-tuned to optimize data preprocessing, prediction, and classification using a dataset from Mansoura University Children's Hospital Diabetes (MUCHD). The system was validated using 548 patient samples, each with 18 significant features. Various metrics such as accuracy, F-score, precision, sensitivity, specificity, and Dice similarity coefficient were used to evaluate the system's performance. The proposed system achieved a remarkable accuracy rate of 99.8%, outperforming existing methods by 0.39%. The study highlights the importance of early diabetes detection and the potential of DL techniques in improving diagnostic accuracy. The method is recommended for predicting diabetes, particularly in pediatric patients, due to its high accuracy and adaptability to different datasets.This study introduces a novel deep learning (DL) technique for early diabetes prediction with high accuracy. The model, based on a multi-layer perceptron (MLP) algorithm within a deep neural network (DNN), consists of ten hidden layers and multiple epochs. The hyperparameters were fine-tuned to optimize data preprocessing, prediction, and classification using a dataset from Mansoura University Children's Hospital Diabetes (MUCHD). The system was validated using 548 patient samples, each with 18 significant features. Various metrics such as accuracy, F-score, precision, sensitivity, specificity, and Dice similarity coefficient were used to evaluate the system's performance. The proposed system achieved a remarkable accuracy rate of 99.8%, outperforming existing methods by 0.39%. The study highlights the importance of early diabetes detection and the potential of DL techniques in improving diagnostic accuracy. The method is recommended for predicting diabetes, particularly in pediatric patients, due to its high accuracy and adaptability to different datasets.