Adaptive systems and methods for seizure detection and confidence indication
Disclosed electroencephalogram (EEG) monitoring and detection systems and methods can detect differentiated electrographic seizure characteristics of different seizure types across a diverse group of patients. Disclosed systems and methods can utilize EEG data collected by discrete wireless EEG sensors positioned on a scalp of a patient.
1. A method for detecting seizure events using electroencephalogram (EEG) signals, the method comprising, by one or more processors:
obtaining one or more EEG data segments collected by a plurality of EEG sensors positioned on a scalp of a patient;
processing each of the one or more EEG data segments to obtain a plurality of probabilities of occurrence of one or more electrographic seizure characteristics;
interpreting, by each of a plurality of event identifiers, the plurality of probabilities to obtain a plurality of event interpretations, each event interpretation reflecting a confidence value that represents a confidence of the interpretation being a true positive event;
selecting an event interpretation from the plurality of event interpretations based on the confidence values; and
outputting a label indicating 1) a seizure event based on the selected event interpretation and 2) a confidence indicator determined based on the confidence value of the selected event interpretation, the confidence indicator reflecting a confidence in the seizure event being a true positive event.
2. The method of claim 1 , wherein the label indicates a start time and duration of a discrete seizure event.
3. The method of claim 1 , wherein the label indicates a prevalence of an electrographic seizure characteristic over a duration of time.
4. The method of claim 1 , wherein outputting the label comprises displaying the seizure event along with the confidence indicator together with the one or more EEG data segments.
5. The method of claim 4 , wherein the one or more EEG data segments are displayed as a longitudinal transverse montage that comprises four channels of EEG data collected by four discrete wireless EEG sensors positioned at a left forehead, at a right forehead, behind a left ear, and behind a right ear of the patient and a plurality of channels derived from at least two channels of the four channels of EEG data.
6. The method of claim 1 , wherein the confidence indicator comprises one of a plurality of confidence value ranges associated with high, moderate, and/or low confidence.
7. The method of claim 1 , wherein selecting the event interpretation from the plurality of event interpretations comprises selecting an event interpretation reflecting the highest confidence value.
8. The method of claim 1 , further comprising:
merging the seizure event with another seizure event into a merged seizure event, wherein outputting the label comprises outputting a merged label indicating 1) the merged seizure event and 2) a merged confidence indicator that reflects a confidence in the merged seizure event being a true positive event.
9. The method of claim 1 , wherein each event identifier combines temporal sequences of the plurality of probabilities to output a presence or absence of the seizure event.
10. The method of claim 1 , wherein the processing comprises processing by one or more classifiers each of the one or more EEG data segments to obtain the plurality of probabilities.
11. A non-transitory computer readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform a method for detecting seizure events using electroencephalogram (EEG) signals, the method comprising:
obtaining one or more EEG data segments collected by a plurality of EEG sensors positioned on a scalp of a patient;
processing each of the one or more EEG data segments to obtain a plurality of probabilities of occurrence of one or more electrographic seizure characteristics;
interpreting, by each of a plurality of event identifiers, the plurality of probabilities to obtain a plurality of event interpretations, each event interpretation reflecting a confidence value that represents a confidence of the interpretation being a true positive event;
selecting an event interpretation from the plurality of event interpretations based on the confidence values; and
outputting a label indicating 1) a seizure event based on the selected event interpretation and 2) a confidence indicator determined based on the confidence value of the selected event interpretation, the confidence indicator reflecting a confidence in the seizure event being a true positive event.
12. The non-transitory computer readable medium of claim 11 , wherein the label indicates a start time and duration of a discrete seizure event.
13. The non-transitory computer readable medium of claim 11 , wherein the label indicates a prevalence of an electrographic seizure characteristic over a duration of time.
14. The non-transitory computer readable medium of claim 11 , wherein outputting the label comprises displaying the seizure event along with the confidence indicator together with the one or more EEG data segments.
15. The non-transitory computer readable medium of claim 14 , wherein the one or more EEG data segments are displayed as a longitudinal transverse montage that comprises four channels of EEG data collected by four discrete wireless EEG sensors positioned at a left forehead, at a right forehead, behind a left ear, and behind a right ear of the patient and a plurality of channels derived from at least two channels of the four channels of EEG data.
16. The non-transitory computer readable medium of claim 11 , wherein the confidence indicator comprises one of a plurality of confidence value ranges associated with high, moderate, and/or low confidence.
17. The non-transitory computer readable medium of claim 11 , wherein selecting the event interpretation from the plurality of event interpretations comprises selecting an event interpretation reflecting the highest confidence.
18. The non-transitory computer readable medium of claim 11 , wherein the method further comprises:
merging the seizure event with another seizure event into a merged seizure event, wherein outputting the label comprises outputting a merged label indicating 1) the merged seizure event and 2) a merged confidence indicator that reflects a confidence in the merged seizure event being a true positive event.
19. The non-transitory computer readable medium of claim 11 , wherein each event identifier is configured to combine temporal sequences of the plurality of probabilities to output a presence or absence of the seizure event.
20. The non-transitory computer readable medium of claim 11 , wherein the processing comprises processing by one or more classifiers each of the one or more EEG data segments to obtain the plurality of probabilities.