Development of A Sequential CNN Model With Three Hidden Layers For Diabetes Prediction

The early diagnosis and treatment of diabetes can contribute to the mitigation of associated risks and effects. Therefore, it is imperative to anticipate and identify the ailment at an early stage through the utilization of dependable procedures that can offer forecasts characterized by a substantial level of dependability and precision. This study utilizes the Diabetes Readmission Dataset, comprising 101,766 records and encompassing 50 features. After selecting the most pertinent features, the dataset is partitioned into a training set and a test set. Subsequently, a sequential model employing deep learning, specifically a convolutional neural network (CNN) with three hidden layers, is constructed for prediction purposes. The correctness of the model was assessed through the utilization of performance testing metrics, resulting in a recorded accuracy rate of 99.53%. The findings of this study have the potential to inform the development of personalized treatment approaches that address the individualized requirements of patients, hence enhancing the quality of healthcare provide.