ABSTRACT
Diabetes is a chronic condition that must be identified and treated early to avoid long-term problems Deep learning models have demonstrated encouraging outcomes in the early detection of diabetes and offering tailored treatment plans. In this research, a deep learning-based method for diabetes illness diagnosis is proposed. The method makes Utilization a dataset of clinical and laboratory data, such as age, blood pressure, blood glucose levels, and body mass index, and genetic variables, that was collected from diabetes patients. To begin with, feature selection techniques are employed to determine which characteristics have the greatest impact on diabetes categorization. The dataset is then used to train and assess distinguishable. Algorithms, including Random Forests, Decision Trees, and Support Vector Machines. The models’ efficacy is evaluated using performance indicators like F1-score, recall, accuracy, and precision. The results indicate that elevated percentages of diabetes detection accuracy can be attained by combining particular features with deep learning methods. The suggested system may help medical practitioners screen patients for diabetes in the early stages, which could result in an early diagnosis and suitable treatment. To validate the system's performance on bigger and more varied datasets, more investigation is needed.
