The abatement of micropollutants by ozonation can be accurately calculated using the exposures of molecular ozone (O₃) and hydroxyl radical (^●OH) (i.e., ∫[O₃]dt and ∫[^●OH]dt). In the actual ozonation process, the ∫[O_​3]dt values can be technically determined by monitoring the O_​3 decay during the process. However, determining ∫[^●OH]dt is nearly impossible in the field, which necessitates developing models to predict ∫[^●OH]dt from measurable parameters. This study demonstrates the development of machine learning models to predict ∫[^●OH]dt (the output variable) from five basic input variables (i.e., pH, dissolved organic carbon concentration, alkalinity, temperature, and O_​3 dose) and two optional ones (i.e., ∫[O_​3]dt and instantaneous ozone demand, IOD). To develop the models, four different machine learning methods (i.e., random forest, support vector regression, artificial neural network, Gaussian process regression) were employed, using input and output variables measured (or determined) in 130 different natural water samples. The results indicated that incorporating ∫[O₃]dt as an input variable evidently improved the accuracy of prediction models, suggesting that ∫[O₃]dt plays a crucial role as a key variable reflecting the OH-yielding characteristics of dissolved organic matter. Conversely, IOD had minimal impact on the accuracy of the prediction models. Generally, machine learning-based prediction models outperformed those based on response surface methodology developed as a control. Notably, models utilizing the Gaussian process regression algorithm demonstrated the highest coefficient of determination among prediction models.