Abstracts

Rationale: Caregiver intervention is the primary method for mitigating seizure-related adverse events but there exists no reliable method for detecting the occurrence of a significant seizure in the non-clinical setting. We sought to measure physiological responses that can arise from changes in autonomic nervous system activity in children with active epilepsy and to identify patterns that correlate with seizures but not with non-seizure behavior. The long term goal is to develop an effective wearable seizure monitoring and alert system that detects the occurrence of a generalized seizure 95% of the time with a false event rate of < 10%. Methods: Three dependent children with drug-resistant epilepsy, age range 2-31, were enrolled into the study after parental consent. 1 child had tonic and tonic-clonic (TC) seizures several times a week, 1 child had myoclonic (MC) seizures several times per day, and one child had multiple seizure types, including tonic, TC, MC and partial seizures, several times per day. All 3 parents were given commercially available noninvasive and unobtrusive sensors that included heart rate, respiration, and torso orientation. One parent was also given a surface electromyography (EMG) sensor. Sensors were attached to the children multiple times for 1-24 hours over several months. Data were collected during over 50 seizure events. Enrollment of additional study subjects is ongoing and we are continuing to apply sophisticated digital signal processing algorithms to select the optimal array of noninvasive physiological sensors to meet the study goals. Electrocardiogram (ECG) patch electrodes are being added to the sensor array in July 2010 to monitor rhythm and detect cardiac arrhythmia associated with seizures. Collection of data on a vEEG unit will commence in the Fall of 2010. Results: Example data from two seizure types are presented in Figure 1. The dotted lines denote the onset of the seizures as defined by caregiver observation. The analysis revealed multiple physiological changes that correlated with each seizure, including a rapid increase in heart rate, a rapid change in respiration rate and depth of breath (combined in the integrated respiration waveform), and a sudden change in torso orientation. Using cardiac parameters alone, our preliminary detection algorithm identified 7 out of 7 generalized tonic-clonic events and 15 out of 16 myoclonic events, for a detection rate of 94%. EMG data also differentiated seizure activity from normal activity. Seizure onset was detected by a direction trend in muscle activity along the muscle fiber (Figure 2). Conclusions: The data suggest that development of an effective monitor to detect generalized seizures in the home setting is possible and that detection of changes in the heart rate and rhythm should be key components. Such a system could have ground-breaking impact on the potential prevention of seizure-related injury, status epilepticus and SUDEP, as well as improvement of quality of life and increased independence for both caregivers and persons with epilepsy.