Aims: Catheter ablation based on spatiotemporal dispersion (STD) has been proposed to treat persistent atrial fibrillation (AF) effectively. STD patterns are thought to be associated with active zones sustaining arrhythmia, and are targets for successful ablation. This study assesses the accuracy of different machine learning models trained on realistic synthetic multipolar electrograms (EGM) for automatically locating STD patterns. Those models are expected to support interventional cardiologists, helping EGM interpretation and guiding catheter ablation procedures. Methods: Using the openCARP cardiac electrophysiology simulation software, realistic 10-leads EGM records are synthesized, simulating the PentaRay multielectrode mapping catheter (Biosense Webster Inc., Irvine, CA, USA). Ionic current and conductivity parameters are set to create fibrosis in the simulated heart tissue, leading to rotor-like propagation patterns linked to STD behavior. A balanced dataset is obtained. Different machine learning methods are considered: random forests, linear models, deep learning techniques such as long short-term memory (LSTM) networks and convolutional neural networks (CNN). We focus on balancing fitting and stability, to get the best accuracy considering an accountable need of computational power and running time. Experiments are performed using Python. Results: A total of 4000 synthetic EGM are included, split into 80% training and 20% test. Precision, recall, their harmonic mean F1 score, and accuracy are considered as key performance indicators (KPIs), for each of the tested classification models. The best performance is obtained by the Random Forest classification method, which classifies correctly 97% of STDs on the training set and 95% on the test set. Conclusion: The Random Forest model is able to identify STD patterns on a realistic synthetic multipolar EGM dataset with a 95% level of accuracy on the test set, balancing fitting and stability, making it a good candidate for further assessment on a real dataset to support interventional cardiologists in the catheter ablation procedure.