Table 1.
Requirement | Test |
---|---|
EPAD shall disconnect when the INS battery reaches 25% to prolong battery life and prevent loss of therapy | Canine subject's INS battery was partially charged, and EPAD disconnection was observed when it reached 25% power |
The embedded LDA detector shall identify at least 80% of physician identified electrographic seizures with a false positive rate of <20% | The benchtop device is attached to electrodes immersed in a saline bath. EEG signals previously recorded from canines with epilepsy were electrically conducted into the saline bath using an Arduino. EPAD recorded the EEG and LDA seizure detections, and these detections were compared to the canine EEG signal annotations |
The Application shall modulate the amplitude and frequency of stimulation in response to the output of iEEG analytics, with frequencies and amplitudes as configured by the physician. iEEG analytics shall identify iEEG characteristics similar to data preceding physician-identified seizures (pre-ictal) | Phase I: The python executable performing seizure prediction was trained on retrospective canine iEEG data and tested on over 60 days of data on a separate computer to verify performance Phase 2: The same executable running on the tablet computer as part of EPAD was trained to identify delta wave sleep and to initiate very low amplitude (0.5 mA) stimulation on a canine subject. Recorded iEEG data was reviewed to confirm stimulation artifact was visible during delta wave sleep |
EPAD shall provide the ability to conduct a stimulation trial, cycling through at least 12 sets of stimulation amplitudes and frequencies on up to 2 sets of electrodes | Stimulation trial was configured with notably different amplitudes and frequencies on different electrodes. The stimulation trial was run first on the benchtop device and then on a canine subject's device with EEG recording enabled. Stimulation artifacts on recording electrodes were used to confirm relative stimulation rate and amplitude changes |