A Hybrid Attention U-Net and K-Means Approach for Multi-View Clustering in Alzheimer’s Disease Prediction

Abstract

Alzheimer's disease (AD) is a catastrophic neurological disorder that progressively impairs memory and cognitive function. Accurate prediction of its progression is critical for early intervention and distinctive treatment. The proposed system outlays the idea of predicting Alzheimer’s disease progression using an attention U-net based multi-view clustering model. Attention U-Net is generally a robust deep learning framework that is efficient in image segmentation and classification tasks and extracts intricate spatial details from medical or industrial imaging datasets that are critical for precise analysis and decision-making.Traditional Alzheimer’s disease progression prediction models depended on single-view data, restricting their capacity to capture the complexities of the disease’s stages. Thus, our model utilizes attention mechanism along with multi-view data to emphasize the salient areas, presents a loss function that combines sparse categorical cross-entropy with focal and dice loss to address the imbalanced datasets, and provides a complete depiction.The results show immense improvement in segmentation accuracy, allowing accurate prediction of AD progression and robust clinical application groundwork. This research paper introduces a new multi-view clustering methodology that combines an Attention U-Net for feature extraction and K-Means clustering for patient grouping according to disease progressionstages Experiments on an Alzheimer’s-related dataset demonstrate that the proposed framework improves cluster coherence and offers insightfulknowledgeaboutthediseaseprogression.Theresultsdemonstrate that the multi-view clustering model is superior to conventional single-view models, and it can be a valuable tool for clinical use.