Cusps, characterized by a series of horns and embayments, are notable coastal features observed along shorelines in various areas. When the horn-to-horn distances of cusps are substantial, they are referred to as mega cusps. There is a scarcity of measured morphological data for mega cusps, especially long-term data, making it challenging to acquire comprehensive insights. In this study, the morphodynamic characteristics of mega cusps such as mega cusp cross-shore extent and length were analyzed using data collected from Hasaki Beach in Japan over the period of 2007 to 2020. The analysis aimed to address the lack of accurate formulation and numerical models for simulating mega cusp morphological characteristics, which poses another obstacle in mega cusp prediction. To overcome this, different Artificial Neural Networks (ANNs) were employed to predict the horn-to-horn and horn-to-embayment distances of mega cusps following storm events. Storm wave parameters, including significant wave height, wave period, and wave direction, along with additional data such as cumulative energy flux and storm event duration, were utilized as input variables for the ANNs. The accuracy of the predicted results was assessed using the Pearson correlation coefficient, which was calculated based on a comparison with field data. The findings of this study demonstrated that the inclusion of significant wave height, wave period, wave direction, cumulative energy flux, and storm duration in the ANNs yielded satisfactory accuracy in predicting mega cusp characteristics. Overall, this investigation contributes to advancing our understanding of beach mega cusp dynamics and offers a viable approach for predicting mega cusp characteristics by utilizing artificial neural networks and incorporating relevant storm wave parameters.