Alzheimer’s Disease (AD) is a neurodegenerative disorder that poses a serious threat to the quality of life of the elderly population, and its early diagnosis is crucial for slowing disease progression. This study proposes a novel multimodal classification framework that integrates structural magnetic resonance imaging (sMRI) and APOE genotype data with deep learning techniques to improve AD classification performance. Specifically, three core modules are designed: 1) the Spatial Enhancement Weighting Module (SEWM), which enhances local anatomical feature extraction from sMRI data using channel and spatial attention mechanisms; 2) the Parallel Multi-Scale Feature Fusion Module (PMSFM), which fuses multi-scale features through pixel-level attention mechanisms to capture subtle pathological changes; and 3) the Abstract Patch (AP) module, which compresses high-dimensional data using depthwise separable convolutions and multi-head self-attention mechanisms to reduce computational complexity. Experiments were conducted using data from the ADNI database, with ADNI1, ADNI GO, and ADNI2 as the training set and ADNI3 as the independent validation set. Results show that the proposed method achieves an accuracy of 86.2\% in the three-class classification task (CN vs. MCI vs. AD) and up to 96.2\% in the binary classification task (CN vs. AD), surpassing existing multimodal models. Further analysis demonstrates that the synergistic effects among the modules effectively enhance the model's ability to distinguish MCI heterogeneity and AD pathological features. This study provides an efficient technical approach for the early diagnosis of AD. Future work will focus on model lightweighting and cross-center data validation.

Medical artificial intelligence systems often rely on a single model, which may give inconsistent answers and does not reflect the team-based nature of clinical consultation. This study develops a collaborative framework in which several agents analyze a medical question from different professional perspectives, discuss the evidence, vote on the proposed answer, and revise unresolved questions up to three times. Each step and stopping condition is recorded in a time-ordered workflow so that the decision process can be reviewed. The framework was evaluated on three regional versions of a medical examination dataset and three medical image question-answering datasets. It achieved an accuracy of 80.1% on the Mainland examination dataset and produced the highest reported results among the evaluated systems on ADAM. Stronger image models performed better on ACRIMA and Covid CT, showing that collaboration remains limited by the underlying visual model. Additional model-based quality scores are reported only as exploratory results because they were not validated by an independent model or human experts. These findings show that structured collaboration can make medical artificial intelligence workflows more transparent.