Diabetes, silent but relentless, is not an isolated threat but a public health crisis affecting millions of people. This metabolic disorder, characterized by high blood glucose levels, triggers serious complications that affect multiple organs in the human body. This condition not only compromises the quality of life of those who suffer from it but also represents a significant burden on healthcare systems, especially when unplanned hospital readmissions occur. These readmissions, especially those occurring within 30 days of discharge, may reflect deficiencies in treatment continuity. Preventing readmissions is a priority because it reduces hospital costs, improves clinical outcomes, and optimizes the use of medical resources. Therefore, the research implemented a predictive model of the risk of hospital readmission in diabetic patients using machine learning algorithms. A five-phase methodology was applied: data set acquisition; preprocessing; feature selection; implementation of machine learning algorithms (DT, LightGBM, RF, XG Boost, Ada Boost, and Gradient Boosting); and model evaluation. The best results were obtained with the RF algorithm, with an accuracy of 95.1%, a predictive accuracy of 99.93%, a recall of 90.37%, an F1 score of 94.91%, and an AUC of 0.9984 on the ROC curve. In conclusion, machine learning algorithms demonstrated high effectiveness in predicting hospital readmissions in diabetic patients, where their use allows for the anticipation of risks and the optimization of medical care. This favors more timely and effective preventive interventions.