Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Geometric-Stochastic Multimodal Deep Learning for Predictive Modeling of SUDEP and Stroke Vulnerability

Preksha Girish, Rachana Mysore, Mahanthesha U, Shrey Kumar, Misbah Fatimah Annigeri, Tanish Jain

TL;DR

This work introduces GSM-DL, a geometric-stochastic multimodal deep learning framework that unifies EEG, ECG, respiration, SpO2, EMG, and fMRI data to predict SUDEP risk and stroke vulnerability. By embedding signals on Lie-group–invariant Riemannian manifolds, modeling long-memory dynamics with fractional stochastic differential equations, enforcing energy-conserving Hamiltonian neural networks, and incorporating cross-modal attention and diffusion over structural brain graphs, the approach yields interpretable biomarkers tied to manifold geometry, memory, and network diffusion. On the MULTI-CLARID dataset, GSM-DL outperforms state-of-the-art baselines in SUDEP and stroke prediction and provides mechanistic biomarkers such as energy entropy and diffusion centrality for risk stratification. The framework offers a principled, interpretable foundation for early detection and personalized risk assessment in neural-autonomic disorders, with potential for clinical translation pending prospective validation.

Abstract

Sudden Unexpected Death in Epilepsy (SUDEP) and acute ischemic stroke are life-threatening conditions involving complex interactions across cortical, brainstem, and autonomic systems. We present a unified geometric-stochastic multimodal deep learning framework that integrates EEG, ECG, respiration, SpO2, EMG, and fMRI signals to model SUDEP and stroke vulnerability. The approach combines Riemannian manifold embeddings, Lie-group invariant feature representations, fractional stochastic dynamics, Hamiltonian energy-flow modeling, and cross-modal attention mechanisms. Stroke propagation is modeled using fractional epidemic diffusion over structural brain graphs. Experiments on the MULTI-CLARID dataset demonstrate improved predictive accuracy and interpretable biomarkers derived from manifold curvature, fractional memory indices, attention entropy, and diffusion centrality. The proposed framework provides a mathematically principled foundation for early detection, risk stratification, and interpretable multimodal modeling in neural-autonomic disorders.

Geometric-Stochastic Multimodal Deep Learning for Predictive Modeling of SUDEP and Stroke Vulnerability

TL;DR

This work introduces GSM-DL, a geometric-stochastic multimodal deep learning framework that unifies EEG, ECG, respiration, SpO2, EMG, and fMRI data to predict SUDEP risk and stroke vulnerability. By embedding signals on Lie-group–invariant Riemannian manifolds, modeling long-memory dynamics with fractional stochastic differential equations, enforcing energy-conserving Hamiltonian neural networks, and incorporating cross-modal attention and diffusion over structural brain graphs, the approach yields interpretable biomarkers tied to manifold geometry, memory, and network diffusion. On the MULTI-CLARID dataset, GSM-DL outperforms state-of-the-art baselines in SUDEP and stroke prediction and provides mechanistic biomarkers such as energy entropy and diffusion centrality for risk stratification. The framework offers a principled, interpretable foundation for early detection and personalized risk assessment in neural-autonomic disorders, with potential for clinical translation pending prospective validation.

Abstract

Sudden Unexpected Death in Epilepsy (SUDEP) and acute ischemic stroke are life-threatening conditions involving complex interactions across cortical, brainstem, and autonomic systems. We present a unified geometric-stochastic multimodal deep learning framework that integrates EEG, ECG, respiration, SpO2, EMG, and fMRI signals to model SUDEP and stroke vulnerability. The approach combines Riemannian manifold embeddings, Lie-group invariant feature representations, fractional stochastic dynamics, Hamiltonian energy-flow modeling, and cross-modal attention mechanisms. Stroke propagation is modeled using fractional epidemic diffusion over structural brain graphs. Experiments on the MULTI-CLARID dataset demonstrate improved predictive accuracy and interpretable biomarkers derived from manifold curvature, fractional memory indices, attention entropy, and diffusion centrality. The proposed framework provides a mathematically principled foundation for early detection, risk stratification, and interpretable multimodal modeling in neural-autonomic disorders.

Paper Structure

This paper contains 35 sections, 37 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Geometric Embeddings and Lie Group Symmetries
  3. Fractional Stochastic Dynamics
  4. Hamiltonian Neural Networks for Energy Flow
  5. Attention-Based Cross-Modality Coupling
  6. Network Diffusion for Stroke Propagation
  7. Related Work
  8. EEG-based Deep Learning for Seizure Detection
  9. Wearable Device and Multimodal Sensor Approaches
  10. SUDEP Risk Assessment and Data Standardization
  11. Geometric and Stochastic Modeling of Physiological Signals
  12. Network Diffusion and Stroke Propagation
  13. Summary and Research Gap
  14. System Architecture
  15. Data Acquisition and Preprocessing
  16. ...and 20 more sections

Figures (3)

  • Figure 1: System architecture of the proposed Geometric–Stochastic Multimodal Deep Learning (GSM-DL) framework.
  • Figure 2: Receiver Operating Characteristic (ROC) curve demonstrating the classification performance of the proposed GSM-DL model for SUDEP and stroke risk prediction.
  • Figure 3: Visualization of Locus Coeruleus showing ischemic vulnerability or relevant anatomical features.