Age-related Macular Degeneration (AMD) is a leading cause of visual impairment and severe vision loss, with an estimated 288 million patients expected by 2040. In recent years, significant advancements in Optical Coherence Tomography (OCT) technology have been made, increasing the professional demands on medical practitioners. Concurrently, the development of deep learning has shown great potential as a powerful tool in diagnosing such diseases. Under this background, the Monitoring Age-related Macular Degeneration Progression In Optical Coherence Tomography (MARIO) project was proposed, which includes two tasks: classify evolution between two pairs of 2-D slices from two consecutive 2D OCT acquisitions and prediction of evolution within 3 months of AMD on OCT 2D slices. This research work proposes two deep learning algorithms to accomplish both tasks. For Task 1, we implemented a Siamese network, and for Task 2, we employed a standalone network architecture combined with a post-processing method. Both network backbones are based on ConvNeXt-large and incorporate a multi-head attention mechanism. Our algorithms demonstrated good performance on the validation set, with an average score of 0.80427 for Task 1 and 0.43726 for Task 2, ranking third during the development phase. The code is available at https://github.com/lumine-1/MARIO_Project.