In order to prevent wellbore instability, wellbore collapse, and even severe safety incidents such as blowouts and overflow caused by casing damage in water injection wells, a precise prediction method was developed to identify casing failure in advance. ISSA (improved sparrow search algorithm) was used to optimize RF (random forest) model for predicting casing damage in injection wells within the following year. ISSA was enhanced by initializing the population using circle chaotic mapping and introducing Gaussian disturbance to improve its global search capability and convergence speed. The optimized ISSA was then employed to select the hyperparameters of RF model, and an ISSA-RF prediction model for casing damage was constructed. The results show that the model achieved a prediction accuracy of 0.92, with recall rates of 0.92, 0.89, and 0.90 for each category, respectively. It is concluded that the proposed method outperforms other comparative models in terms of prediction accuracy and recall, demonstrating its effectiveness and advancement. This study provides important technical support for casing damage prediction in oilfield water injection wells.