SDE-Driven Spatio-Temporal Hypergraph Neural Networks for Irregular Longitudinal fMRI Connectome Modeling in Alzheimer's Disease
arXiv:2603.20452v1 Announce Type: new Abstract: Longitudinal neuroimaging is essential for modeling disease progression in Alzheimer's disease (AD), yet irregular sampling and missing visits pose substantial challenges for learning reliable temporal representations. To address this challenge, we propose SDE-HGNN, a stochastic differential equation (SDE)-driven spatio-temporal hypergraph neural network for irregular longitudinal fMRI connectome modeling. The framework first employs an SDE-based reconstruction module to recover continuous latent trajectories from irregular observations. Based on these reconstructed representations, dynamic hypergraphs are constructed to capture higher-order interactions among brain regions over time. To further model temporal evolution, hypergraph convolution parameters evolve through SDE-controlled recurrent dynamics conditioned on inter-scan intervals, enabling disease-stage-adaptive connectivity modeling. We also incorporate a sparsity-based importan
arXiv:2603.20452v1 Announce Type: new Abstract: Longitudinal neuroimaging is essential for modeling disease progression in Alzheimer's disease (AD), yet irregular sampling and missing visits pose substantial challenges for learning reliable temporal representations. To address this challenge, we propose SDE-HGNN, a stochastic differential equation (SDE)-driven spatio-temporal hypergraph neural network for irregular longitudinal fMRI connectome modeling. The framework first employs an SDE-based reconstruction module to recover continuous latent trajectories from irregular observations. Based on these reconstructed representations, dynamic hypergraphs are constructed to capture higher-order interactions among brain regions over time. To further model temporal evolution, hypergraph convolution parameters evolve through SDE-controlled recurrent dynamics conditioned on inter-scan intervals, enabling disease-stage-adaptive connectivity modeling. We also incorporate a sparsity-based importance learning mechanism to identify salient brain regions and discriminative connectivity patterns. Extensive experiments on the OASIS-3 and ADNI cohorts demonstrate consistent improvements over state-of-the-art graph and hypergraph baselines in AD progression prediction. The source code is available at https://anonymous.4open.science/r/SDE-HGNN-017F.
Executive Summary
This article proposes SDE-HGNN, a spatio-temporal hypergraph neural network for modeling irregular longitudinal fMRI connectomes in Alzheimer's disease. The framework involves an SDE-based reconstruction module, dynamic hypergraphs, and SDE-controlled recurrent dynamics to capture disease-stage-adaptive connectivity. A sparsity-based importance learning mechanism identifies salient brain regions and discriminative connectivity patterns. Experiments demonstrate consistent improvements over state-of-the-art graph and hypergraph baselines in AD progression prediction. The proposed method addresses a critical challenge in longitudinal neuroimaging and has the potential to improve AD diagnosis and treatment. However, further validation and comparison with other machine learning approaches are necessary.
Key Points
- ▸ SDE-HGNN addresses irregular longitudinal fMRI sampling and missing visits challenges
- ▸ The framework uses SDE-based reconstruction, dynamic hypergraphs, and SDE-controlled recurrent dynamics
- ▸ Experiments demonstrate improvements over state-of-the-art graph and hypergraph baselines in AD progression prediction
Merits
Strength
The proposed framework addresses a critical challenge in longitudinal neuroimaging and has the potential to improve AD diagnosis and treatment.
Demerits
Limitation
Further validation and comparison with other machine learning approaches are necessary to confirm the generalizability and robustness of SDE-HGNN.
Expert Commentary
The proposed SDE-HGNN framework is a significant contribution to the field of neuroimaging and machine learning. The use of SDE-based reconstruction and dynamic hypergraphs is innovative and addresses a critical challenge in longitudinal neuroimaging. However, further validation and comparison with other machine learning approaches are necessary to confirm the generalizability and robustness of SDE-HGNN. The article's focus on AD progression prediction is timely and relevant, and the potential implications for AD diagnosis and treatment are substantial. Overall, this article is a strong contribution to the field and has the potential to influence future research and clinical practice.
Recommendations
- ✓ Future research should focus on validating SDE-HGNN on larger and more diverse datasets to confirm its generalizability and robustness
- ✓ Comparison with other machine learning approaches should be conducted to confirm the superiority of SDE-HGNN in AD progression prediction
Sources
Original: arXiv - cs.LG
