SOUL: An Energy-Efficient Unsupervised Online Learning Seizure Detection Classifier

Implantable devices that record neural activity and detect seizures have been adopted to issue warnings or trigger neurostimulation to suppress epileptic seizures. Typical seizure detection systems rely on high-accuracy offline-trained machine learning classifiers that require manual retraining when seizure patterns change over long periods of time. For an implantable seizure detection system, a low-power, at-the-edge, online learning algorithm can be employed to dynamically adapt to the neural signal drifts, thereby maintaining high accuracy without external intervention. This work proposes SOUL: Stochastic-gradient-descent-based Online Unsupervised Logistic regression classifier. After an initial offline training phase, continuous online unsupervised classifier updates are applied in situ, which improves sensitivity in patients with drifting seizure features. SOUL was tested on two human electroencephalography (EEG) datasets: the Children’s Hospital Boston and the Massachusetts Institute of Technology (CHB-MIT) scalp EEG dataset and a long ( $\!>$ 100 h) intracranial EEG dataset. It was able to achieve an average sensitivity of 97.5% and 97.9% for the two datasets, respectively, at >95% specificity. Sensitivity improved by at most 8.2% on long-term data when compared to a typical seizure detection classifier. SOUL was fabricated in Taiwan Semiconductor Manufacturing Company (TSMC’s) 28 nm process occupying 0.1 mm2 and achieves 1.5 nJ/classification energy efficiency, which is at least $24\times $ more efficient than state-of-the-art.

[1]  Jerald Yoo,et al.  A One-Shot Learning, Online-Tuning, Closed-Loop Epilepsy Management SoC with 0.97μJ/Classification and 97.8% Vector-Based Sensitivity , 2021, 2021 Symposium on VLSI Circuits.

[2]  Rikky Muller,et al.  A 1.5nJ/cls Unsupervised Online Learning Classifier for Seizure Detection , 2021, 2021 Symposium on VLSI Circuits.

[3]  Ning Wang,et al.  4.5 BioAIP: A Reconfigurable Biomedical AI Processor with Adaptive Learning for Versatile Intelligent Health Monitoring , 2021, 2021 IEEE International Solid- State Circuits Conference (ISSCC).

[4]  Michael I. Jordan,et al.  Unsupervised Online Learning for Long-Term High Sensitivity Seizure Detection , 2020, 2020 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC).

[5]  Quan Sun,et al.  26.3 A Closed-Loop Neuromodulation Chipset with 2-Level Classification Achieving 1.5Vpp CM Interference Tolerance, 35dB Stimulation Artifact Rejection in 0.5ms and 97.8% Sensitivity Seizure Detection , 2020, 2020 IEEE International Solid- State Circuits Conference - (ISSCC).

[6]  Roman Genov,et al.  26.2 A Neuromorphic Multiplier-Less Bit-Serial Weight-Memory-Optimized 1024-Tree Brain-State Classifier and Neuromodulation SoC with an 8-Channel Noise-Shaping SAR ADC Array , 2020, 2020 IEEE International Solid- State Circuits Conference - (ISSCC).

[7]  Chia-Hsiang Yang,et al.  A 1.9-mW SVM Processor With On-Chip Active Learning for Epileptic Seizure Control , 2020, IEEE Journal of Solid-State Circuits.

[8]  Lili Chen,et al.  Automatic Diagnosis of Epileptic Seizure in Electroencephalography Signals Using Nonlinear Dynamics Features , 2019, IEEE Access.

[9]  Brian Litt,et al.  Epilepsyecosystem.org: crowd-sourcing reproducible seizure prediction with long-term human intracranial EEG , 2018, Brain : a journal of neurology.

[10]  Mahsa Shoaran,et al.  Energy-Efficient Classification for Resource-Constrained Biomedical Applications , 2018, IEEE Journal on Emerging and Selected Topics in Circuits and Systems.

[11]  Ke Li,et al.  Epileptic Seizure Classification of EEGs Using Time–Frequency Analysis Based Multiscale Radial Basis Functions , 2018, IEEE Journal of Biomedical and Health Informatics.

[12]  Roman Genov,et al.  A recursive-memory brain-state classifier with 32-channel track-and-zoom Δ2 Σ ADCs and Charge-Balanced Programmable Waveform Neurostimulators , 2018, 2018 IEEE International Solid - State Circuits Conference - (ISSCC).

[13]  Levin Kuhlmann,et al.  The circadian profile of epilepsy improves seizure forecasting , 2017, Brain : a journal of neurology.

[14]  Chen Zhang,et al.  Design and Implementation of an On-Chip Patient-Specific Closed-Loop Seizure Onset and Termination Detection System , 2016, IEEE Journal of Biomedical and Health Informatics.

[15]  Tim Oates,et al.  A Flexible Multichannel EEG Feature Extractor and Classifier for Seizure Detection , 2015, IEEE Transactions on Circuits and Systems II: Express Briefs.

[16]  David M. Himes,et al.  Prediction of seizure likelihood with a long-term, implanted seizure advisory system in patients with drug-resistant epilepsy: a first-in-man study , 2013, The Lancet Neurology.

[17]  Anantha Chandrakasan,et al.  An 8-Channel Scalable EEG Acquisition SoC With Patient-Specific Seizure Classification and Recording Processor , 2013, IEEE Journal of Solid-State Circuits.

[18]  Lojini Logesparan,et al.  Optimal features for online seizure detection , 2012, Medical & Biological Engineering & Computing.

[19]  B. Frémont,et al.  SUMMARY OF SAFETY AND EFFECTIVENESS DATA , 2002 .

[20]  Jeffrey M. Hausdorff,et al.  Physionet: Components of a New Research Resource for Complex Physiologic Signals". Circu-lation Vol , 2000 .

[21]  C. Dolea,et al.  World Health Organization , 1949, International Organization.

[22]  J. Kiefer,et al.  Stochastic Estimation of the Maximum of a Regression Function , 1952 .