A Longitudinal Case Study on Detecting Emotional Dysregulation in a Child with Autism Spectrum Disorder Using Wearable Sensors

Emotional dysregulation is a common challenge in children with autism spectrum disorder (ASD), often present as sudden tantrums or mood swings. Continuous monitoring of physiological signals offers a nonintrusive way to understand and potentially anticipate these episodes. This paper presents a work-in-progress case study that provides a longitudinal analysis of three months of multimodal physiological data collected from a 10-year-old child diagnosed with ASD using the Empatica EmbracePlus wearable device. Key signals were analyzed, including electrodermal activity (EDA), heart rate (HR), and accelerometry, to identify patterns linked to emotional dysregulation. To validate events, the caregiver installed video cameras throughout the home and reviewed footage whenever the wearable system flagged abnormal physiological activity. This retrospective confirmation enabled accurate annotation of heightened arousal episodes and supported the use of the system as a real-time assistive tool for a working parent to monitor their child's emotional state. Temporal and arousal-based features were extracted using rolling-window and peak-detection methods. Unsupervised anomaly detection Algorithms were employed to differentiate heightened arousal states from baseline behavior. Results revealed consistent physiological signatures, particularly in EDA and HR, that preceded periods of elevated arousal by several minutes.