Cui, Jie

Objective: To evaluate the feasibility and effectiveness of wrist-worn wearable devices combined with machine learning (ML) approaches for detecting a diverse array of seizure types beyond generalized tonic–clonic (GTC), including focal, generalized, and subclinical seizures. Materials and Methods: Twenty-eight patients undergoing inpatient video-EEG monitoring at Mayo Clinic were concurrently monitored using Empatica E4 wrist-worn devices. These devices captured accelerometry, blood volume pulse, electrodermal activity, skin temperature, and heart rate. Seizures were annotated by neurologists. The data were preprocessed to experiment with various segment lengths (10 s and 60 s) and multiple feature sets. Three ML strategies, XGBoost, deep learning models (LSTM, CNN, Transformer), and ROCKET, were evaluated using leave-one-patient-out cross-validation. Performance was assessed using area under the receiver operating characteristic curve (AUROC), seizure-wise recall (SW-Recall), and false alarms per hour (FA/h). Results: Detection performance varied by seizure type and model. GTC seizures were detected most reliably (AUROC = 0.86, SW-Recall = 0.81, FA/h = 3.03). Hyperkinetic and tonic seizures showed high SW-Recall but also high FA/h. Subclinical and aware-dyscognitive seizures exhibited the lowest SW-Recall and highest FA/h. MultiROCKET and XGBoost performed best overall, though no single model was optimal for all seizure types. Longer segments (60 s) generally reduced FA/h. Feature set effectiveness varied, with multi-biosignal sets improving performance across seizure types. Conclusions: Wrist-worn wearables combined with ML can extend seizure detection beyond GTC seizures, though performance remains limited for non-motor types. Optimizing model selection, feature sets, and segment lengths, and minimizing false alarms, are key to clinical utility and real-world adoption.

Abstract Temporal lobe epilepsy is a common neurological disease characterized by recurrent seizures that often originate within limbic networks involving amygdala and hippocampus. The limbic network is involved in crucial physiologic functions involving memory, emotion and sleep. Temporal lobe epilepsy is frequently drug-resistant, and people often experience comorbidities related to memory, mood and sleep. Deep brain stimulation targeting the anterior nucleus of the thalamus (ANT-DBS) is an established therapy for temporal lobe epilepsy. However, the optimal stimulation parameters and their impact on memory, mood and sleep comorbidities remain unclear. We used an investigational brain sensing-stimulation implanted device to accurately track seizures, interictal epileptiform spikes (IES), and memory, mood and sleep comorbidities in five ambulatory subjects. Wireless streaming of limbic network local field potentials (LFPs) and subject behaviour were captured on a mobile device integrated with a cloud environment. Automated algorithms applied to the continuous LFPs were used to accurately cataloged seizures, IES and sleep-wake brain state. Memory and mood assessments were remotely administered to densely sample cognitive and behavioural response during ANT-DBS in ambulatory subjects living in their natural home environment. We evaluated the effect of continuous low-frequency and duty cycle high-frequency ANT-DBS on epileptiform activity and memory, mood and sleep comorbidities. Both low-frequency and high-frequency ANT-DBS paradigms reduced seizures. However, continuous low-frequency ANT-DBS showed greater reductions in IES, electrographic seizures and better sleep and memory outcomes. These results highlight the potential of synchronized brain sensing and dense behavioural tracking during ANT-DBS for optimizing neuromodulation therapy. While studies with larger patient numbers are needed to validate the benefits of low-frequency ANT-DBS, these findings are potentially translatable to individuals currently implanted with ANT-DBS systems.