To address the challenge of early diagnosis of Alzheimer's disease (AD), this study proposes the 3D-SEConvNeXt model. This model combines ConvNeXt with 3D convolution and introduces the 3D-SE attention module for early AD classification tasks. The experimental data is obtained from the publicly available ADNI database, and the raw Magnetic Resonance Imaging (MRI) data is preprocessed using SPM12. The preprocessed data is then fed into the 3D-SEConvNeXt network for four classification tasks: AD/NC, MCI/NC, AD/MCI, and AD/MCI/NC. The performance of the proposed model is compared with other AD classification models. The experimental results demonstrate that the 3D-SEConvNeXt model consistently outperforms the other models in terms of accuracy, achieving excellent results in early AD diagnosis tasks.