Method and apparatus for detecting seizures
A method and apparatus for detecting seizures with motor manifestations including detecting EMG signals, isolating from the EMG signals spectral data for a plurality of frequency bands, and calculating a T-squared value there from. The T-squared values may be detected in real time, such as in a patient's home environment, and the T-squared data may be compared to a threshold T-squared value to determine whether an alarm is sent.
1. A method of detecting seizures with motor manifestations comprising:
detecting electromyography signals for a time period, the time period comprising a reference period and a monitoring period;
using digital filtering to isolate from the electromyography signals spectral data for a plurality of frequency bands selected from the range of about 2 Hz to about 1000 Hz;
calculating a first principal component analysis value, the first principal component analysis value being determined from spectral data for the plurality of frequency bands from at least one part of the reference period;
calculating a second principal component analysis value, the second principal component analysis value being determined from spectral data for the plurality of frequency bands from at least one part of the monitoring period;
comparing the second principal component analysis value to the first principal component analysis value to obtain a comparison; and
determining whether to trigger an alarm condition using said comparison of the first principal component analysis value to the second principal component analysis value.
2. The method of claim 1 wherein the comparison of the second principal component analysis value to the first principal component analysis value includes calculating a number of standard deviations in which the second principal component analysis value differs from the first principal component analysis value; and
wherein said determining whether to trigger an alarm comprises calculating if said number of standard deviations exceeds a threshold number of standard deviations.
3. The method of claim 1 , wherein the plurality of frequency bands comprises a first frequency band from about 300 Hz to about 400 Hz, a second frequency band from about 130 Hz to about 240 Hz, and a third frequency band from about 30 Hz to about 40 Hz.
4. The method of claim 2 , wherein the alarm condition is not triggered until said number of standard deviations exceeds said threshold number of standard deviations for a set delay time period.
5. The method of claim 4 , wherein the set delay time period is 2 or more seconds.
6. The method of claim 1 , wherein the first and second principal component analysis values are based on the first two principal components.
7. The method of claim 1 wherein the comparison of the second principal component analysis value to the first principal component analysis value includes calculating the difference between said second principal component analysis value and said first principal component analysis value;
wherein the determining of whether to trigger an alarm comprises a calculation of whether said difference is above a threshold principal component analysis value.
8. The method of claim 7 further comprising collecting electromyography signals while a patient is executing a maximum voluntary contraction and measuring a principal component analysis value; and
setting said threshold principal component analysis value based on the principal component analysis value measured during the maximum voluntary contraction.
9. The method of claim 8 wherein the threshold principal component analysis value is greater than the principal component analysis value measured during the maximum voluntary contraction.