Predicting Diabetes Onset Through Symptom Analysis: A Machine-Learning Approach

Around the world, diabetes is a common chronic (long-lasting) disease. One of the best approaches to analyzing early-stage symptoms is machine learning. This paper proposed an approach to predict diabetes type 2 form early stages using a machine-learning model. The study utilizes the ‘Early-stage diabetes risk prediction’ dataset, which includes comprehensive clinical information from individuals at risk of developing diabetes. A Random Forest Classifier is employed to construct a predictive model, achieving an unprecedented accuracy of 98.71% in classifying diabetes onset. The proposed model outperforms other machine-learning algorithms, showcasing its efficacy in identifying potential diabetes cases at an early stage. The model is lightweight compared to previous studies which combined multiple datasets. This makes it useful for portable implementations such as mobile and web applications. The data used by the model can be provided by an individual without the need of a physician or medical equipment; making this approach beneficial for promoting awareness, persuading early detection and prevention to the general public.