Seizure quantification in sunflower syndrome by a wrist-worn device

Sunflower syndrome is a rare photosensitive childhood-onset epilepsy, featuring repetitive handwaving events (HWE) triggered by light. Objective documentation of these HWE can be difficult due to the numerous events occurring daily and/or caregivers who document the seizures but are not always present. Hence, seizure diaries can be underreporting.

We performed a feasibility study in three Belgian Sunflower syndrome individuals to assess the possibility to quantify the HWE by a wrist-worn wearable device (Axivity AX6). We conducted a structured exercise aiming to capture patterns of possible confounders in a controlled environment. Subsequently, patients wore the device for three to six consecutive days and nights at home. Spectral power analyses were performed to characterize the frequency signature of the different movements.

The HWE of patient A and B showed homogeneity and narrow-band frequencies. Patient C did not experience any HWE at the start of the study due to proper seizure control. Regarding HWE, there was a higher spectral power for Gyroscope Z (Gz) compared to Gy. The inter-subject variability for HWE frequency peaks was in the 3-6 Hz range. Computer analysis by visual annotation, without checking the seizure diary, detected 71% of the HWE if the HWE lasted for longer than 5 s (sensitivity 64%). For shorter HWE duration, the detection rate was 50% but seemed to be higher if there was a concordant change of eye movement (63%) (sensitivity 36%). The most obvious confounder was toothbrushing (TB). However, TB showed a different pattern: that is, higher or comparable spectral power for Gy compared to Gz. There was also a higher or comparable spectral power for Gy compared to Gz for "waving hello".

We show that the wearable movement sensor Axivity AX6 can detect HWE in Sunflower syndrome individuals and distinguish them from confounders in a real-world setting.