Epilepsy is a neurological disorder that affects millions of people worldwide and causes severe suffering. By providing opportunities for early intervention and seizure management, early detection of seizures can significantly enhance the quality of life for epileptic patients. Over the past few decades, significant efforts have been made to explore non-invasive methodologies for predicting seizures. Recent advancements have highlighted the potential of Electrocardiogram (ECG), particularly Heart Rate Variability (HRV) analysis, as a valuable biomarker for seizure prediction. However, the effectiveness of these approaches can vary, making it difficult to select the most appropriate strategy. Unlike previous reviews that have mostly focused on methods related to HRV analysis from medical perspectives, we aim to provide a comprehensive review of the machine learning techniques that have been applied to ECG data for predicting epileptic seizure attacks. In this study, we explore the relationship between the cardiovascular system and seizure activity, including the physiological effects of epileptic seizures and their implications for predictive modeling. Additionally, we provide a detailed comparison of available seizure prediction techniques, including a review of public datasets, common methodologies, key components, and evaluation metrics. Finally, the study highlights the strengths and limitations of various approaches while discussing existing challenges and future opportunities. We believe our work lays a basis for developing more sophisticated methods on utilizing the ECG signal for better seizure prediction.