Authors :
Presenting Author: Fatemeh Mohammad Alizadeh Chafjiri, MD – Boston Children's Hospital
Tanuj Hasija, PhD, MSc – Paderborn University, Paderborn, Germany
Emily Peter, BA – Western University of Health Sciences
Michele Jackson, BA – Boston Children's Hospital, 300 Longwood Ave, Boston, MA 02115, USA
Stephanie Dailey, BA – Boston Children's Hospital, 300 Longwood Ave, Boston, MA 02115, USA
Paulina Moehrle, MD Candidate – Boston Children's Hospital, 300 Longwood Ave, Boston, MA 02115, USA
Maurice Kuschel, MSc – Paderborn University, Paderborn, Germany
Xingyan Lui, BS – Boston Children's Hospital
Olivia Mezheritsky, BA – Boston Children's Hospital
Lillian Voke, BS – UMass Chan Medical School
Solveig Vieluf, PhD – LMU University Hospital
Tobias Loddenkemper, MD – Boston Children's Hospital, 300 Longwood Ave, Boston, MA 02115, USA
Rationale:
Anti-seizure medications (ASMs) may affect autonomic nervous system (ANS) activity in patients with epilepsy (PWE). Wearable devices can detect ANS signals, enabling continuous monitoring and personalized treatment. This study analyzed wearable ANS data from patients receiving varying ASM doses.
Methods:
We included pediatric PWE admitted for long-term video-EEG monitoring at Boston Children’s Hospital (2/2015–2/2021) who received ≥1 reduced-schedule ASM before or during pre-surgical evaluation and wore at least one Empatica E4 device on the wrist/ankle.
Exclusion criteria were neurostimulators, Adrenocorticotropin or corticosteroid therapy, continuous ASM infusion, lack of recordings 1 hour before ASM intake (baseline) or within 30 minutes before/after ASM peak (peak window), and seizures during baseline or peak window. Patients were categorized into 3 groups based on ASM intake during the recording: high-dose (regular schedule), low-dose (reduced), and no-dose (none).
We set a signal quality threshold at 0.6 and used the recording with the highest score if two wearables were placed. Only evening (5 pm to < 3 am) recordings were analyzed due to limited daytime recordings. We calculated mean electrodermal activity (EDA), heart rate (HR), peripheral body temperature (TEMP), and derived respiratory rate (RR) from blood volume pulse via spectral analysis. We ran analyses for high-to-no dose and high-to-low dose days separately.
We calculated a repeated-measure analysis of variance (2 dose by 2 time points). We analyzed bimodal signal interactions using a joint principal component analysis (PCA) and canonical correlation analysis (CCA) method for baseline and peak windows. PCA-CCA estimated statistically significant canonical correlations for a pair of modalities, which were combined to obtain a bimodal interaction coefficient (BIC). We computed BIC for all six modality pairs HR-EDA, HR-TEMP, HR-RR, EDA-TEMP, EDA-RR, TEMP-RR to obtain the overall multimodal correlation (normalized between 0 and 1).
Results:
We included 52 patients (42.3% female, median age 12.8 yrs); 34 had high vs low-dose ASM days, 24 had high vs no-dose ASM days (6 overlapped with all three dose days). In the high vs. low-dose group, significant time effects were observed for EDA and HR: EDA increased from baseline to peak, F(1,33) = 9.917, p = 0.003, partial η² = 0.23, while HR decreased, F(1,33) = 4.776, p = 0.036, partial η² = 0.126. No significant time effects occurred in the high vs. no-dose group. No main effects of dose or dose-time interactions were found for any measures. TEMP and RR showed no significant changes (Figure 1). CCA showed increased multimodal correlation from baseline to peak on high-dose days, a decrease on no-dose days, and a slight increase on low-dose days (Figure 2).
Conclusions:
ASMs do not systematically alter peripheral ANS activity, but they do influence multimodal correlations, likely through central mechanisms. These correlation changes may reflect seizure susceptibility, highlighting the potential value of wearable ANS monitoring in clinical practice, for seizure monitoring, medication compliance, and effectiveness tracking.
Funding:
This study was funded by the Epilepsy Research Fund.