MT-NAM: An Efficient and Adaptive Model for Epileptic Seizure Detection

2025-03-11

AI Summary

To address the slow inference speed and underutilized structural priors of existing epilepsy seizure detection models, this paper proposes the Micro-Tree Neural Additive Model (MT-NAM), a lightweight and distillable architecture. Methodologically, we introduce the first micro-tree structure distillation strategy for compressing Neural Additive Models (NAMs) and integrate Test-Time Template Adaptation (T3A) to enable dynamic correction of signal drift. Evaluated on the CHB-MIT scalp EEG dataset, MT-NAM achieves 85.3% window-level sensitivity and 95% specificity, with a 50× speedup in inference latency while recovering the sensitivity of standard NAMs. Our key contributions are: (1) the first distillable micro-tree NAM architecture designed explicitly for real-time epilepsy detection; (2) a novel online adaptive paradigm synergizing structural-aware distillation and T3A; and (3) a significant balance between accuracy and efficiency, enabling low-latency closed-loop neuromodulation applications.

Technology Category

• Machine Learning: Hardware-aware ML
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Enhancing the accuracy and efficiency of machine learning algorithms employed in neural interface systems is crucial for advancing next-generation intelligent therapeutic devices. However, current systems often utilize basic machine learning models that do not fully exploit the natural structure of brain signals. Additionally, existing learning models used for neural signal processing often demonstrate low speed and efficiency during inference. To address these challenges, this study introduces Micro Tree-based NAM (MT-NAM), a distilled model based on the recently proposed Neural Additive Models (NAM). The MT-NAM achieves a remarkable 100$ imes$ improvement in inference speed compared to standard NAM, without compromising accuracy. We evaluate our approach on the CHB-MIT scalp EEG dataset, which includes recordings from 24 patients with varying numbers of sessions and seizures. NAM achieves an 85.3% window-based sensitivity and 95% specificity. Interestingly, our proposed MT-NAM shows only a 2% reduction in sensitivity compared to the original NAM. To regain this sensitivity, we utilize a test-time template adjuster (T3A) as an update mechanism, enabling our model to achieve higher sensitivity during test time by accommodating transient shifts in neural signals. With this online update approach, MT-NAM achieves the same sensitivity as the standard NAM while achieving approximately 50$ imes$ acceleration in inference speed.

Problem

• Improves epileptic seizure detection accuracy and efficiency.
• Addresses low speed and inefficiency in neural signal processing.
• Enhances inference speed without compromising detection sensitivity.

Innovation

• MT-NAM: distilled model for seizure detection
• 100x faster inference speed than standard NAM
• Test-time template adjuster enhances sensitivity