Seizure prediction machine learning models
Various embodiments of the present disclosure provide methods, apparatuses, systems, computing devices, computing entities, and/or the like for detecting and preventing seizure events in real-time. For example, various embodiments provide techniques for detecting and preventing seizure events in real-time that use seizure prediction machine learning models.
1. A computer-implemented method comprising:
receiving, by one or more processors and using a wearable device, an event data object comprising image sensor data indicative of (i) a movement of a face of a user and (ii) a field of view of the user;
inputting, by the one or more processors and to a seizure prediction machine learning model, the event data object;
obtaining, by the one or more processors and from the seizure prediction machine learning model, a seizure event prediction data object, wherein the seizure prediction machine learning model is configured to:
generate one or more predefined facial movement features based at least in part on the image sensor data indicative of the movement of the face of the user from the event data object;
generate a seizure event type that is determined based at least in part on the one or more predefined facial movement features;
generate a predefined seizure triggering condition based at least in part on the image sensor data indicative of the field of view of the user from the event data object; and
generate the seizure event prediction data object based at least in part on the seizure event type and the predefined seizure triggering condition; and
modifying, by the one or more processors and using the wearable device, an opacity of a transparent surface of the wearable device based at least in part on the seizure event prediction data object.
2. The computer-implemented method of claim 1 , wherein the event data object further comprises sensor data indicative of a movement of a head of the user, and the computer-implemented method further comprises:
identifying one or more predefined head movement features from the event data object, wherein:
(i) the seizure event prediction data object is determined based at least in part on the one or more predefined head movement features,
(ii) the seizure prediction machine learning model comprises a first sub-model that is configured to identify the predefined seizure triggering condition in the field of view of the user, and a second sub-model that is configured to determine the one or more predefined head movement features and the one or more predefined facial movement features, and
(iii) the seizure event prediction data object is determined based at least in part on the predefined seizure triggering condition, the one or more predefined head movement features, the one or more predefined facial movement features, and the seizure event type.
3. The computer-implemented method of claim 2 , wherein the second sub-model is configured to determine the one or more predefined head movement features and the one or more predefined facial movement features based at least in part on a Eulerian video processing technique.
4. The computer-implemented method of claim 1 , further comprising:
providing, by the one or more processors, an indication of the seizure event prediction data object to a user computing entity; and
storing, by the one or more processors, information associated with the seizure event prediction data object to a user profile.
5. The computer-implemented method of claim 2 , wherein the sensor data is captured in real-time by an inertial measurement unit (IMU) of the wearable device.
6. The computer-implemented method of claim 5 , wherein the wearable device comprises:
an image sensor,
the IMU, and
a controller configured to:
monitor, via the image sensor, the field of view of the user;
detect the predefined seizure triggering condition in the field of view of the user;
responsive to identifying the predefined seizure triggering condition, generate a seizure event prediction; and
in an instance in which the seizure event prediction satisfies a seizure event prediction threshold, cause initiation of a seizure response protocol associated with the seizure event prediction.
7. The computer-implemented method of claim 1 , wherein modifying the opacity of the transparent surface of the wearable device comprises:
modifying the opacity of at least one portion of a first and/or a second transparent surface of the wearable device.
8. The computer-implemented method of claim 1 , further comprising:
responsive to detecting that at least one predefined seizure triggering condition has ceased, remodifying the opacity of the transparent surface of the wearable device.
9. The computer-implemented method of claim 1 , wherein the one or more predefined facial movement features comprise at least one of rhythmic eye movements, a tasting gesture, or a pupil size.
10. The computer-implemented method of claim 2 , wherein the one or more predefined head movement features comprise three-dimensional head rotations and accelerations.
11. The computer-implemented method of claim 1 , wherein the transparent surface of the wearable device comprises a polymer dispersed liquid crystal (PDLC) film.
12. The computer-implemented method of claim 6 , wherein the seizure response protocol comprises at least one of: (i) providing an audio alert, (ii) transmitting a notification to a computing entity in electronic communication with the wearable device, (iii) obtaining additional sensor data, or (iv) generating a report of the seizure event prediction data object.
13. The computer-implemented method of claim 6 , wherein the controller of the wearable device is further configured to:
store at least a portion of the sensor data associated with the seizure event prediction data object; and
transmit the at least a portion of sensor data to a computing entity that is in electronic communication with the wearable device.
14. The computer-implemented method of claim 1 , wherein the seizure prediction machine learning model comprises a trained neural network machine learning model.
15. A system comprising memory and one or more processors communicatively coupled with the memory, the one or more processors configured to:
receive, using a wearable device, an event data object comprising image sensor data indicative of (i) a movement of a face of a user and (ii) a field of view of the user;
input, to a seizure prediction machine learning model, the event data object;
obtain, from the seizure prediction machine learning model, a seizure event prediction data object, wherein the seizure prediction machine learning model is configured to:
generate one or more predefined facial movement features based at least in part on the image sensor data indicative of the movement of the face of the user from the event data object;
generate a seizure event type that is determined based at least in part on the one or more predefined facial movement features;
generate a predefined seizure triggering condition based at least in part on the image sensor data indicative of the field of view of the user from the event data object; and
generate the seizure event prediction data object based at least in part on the seizure event type and the predefined seizure triggering condition; and
modify, using the wearable device, an opacity of a transparent surface of the wearable device based at least in part on the seizure event prediction data object.
16. The system of claim 15 , wherein the event data object further comprises sensor data indicative of a movement of a head of a user, and the one or more processors are further caused to:
identify one or more predefined head movement features from the event data object, wherein:
(i) the seizure event prediction data object is determined based at least in part on the one or more predefined head movement features,
(ii) the seizure prediction machine learning model comprises a first sub-model that is configured to identify the predefined seizure triggering condition in the field of view of the user, and a second sub-model that is configured to determine the one or more predefined head movement features and the one or more predefined facial movement features, and
(iii) the seizure event prediction data object is determined based at least in part on the predefined seizure triggering condition, the one or more predefined head movement features, the one or more predefined facial movement features, and the seizure event type.
17. One or more non-transitory computer-readable storage media comprising instructions that, when executed by one or more processors, cause the one or more processors to:
receive, from a wearable device, an event data object comprising image sensor data indicative of (i) a movement of a face of a user and (ii) a field of view of the user;
input, to a seizure prediction machine learning model, the event data object;
obtain, from the seizure prediction machine learning model, a seizure event prediction data object, wherein the seizure prediction machine learning model is configured to:
generate one or more predefined facial movement features based at least in part on the image sensor data indicative of the movement of the face of the user from the event data object;
generate, a seizure event type that is determined based at least in part on the one or more predefined facial movement features;
generate a predefined seizure triggering condition based at least in part on the image sensor data indicative of the field of view of the user from the event data object; and
generate the seizure event prediction data object based at least in part on the seizure event type and the predefined seizure triggering condition; and
modify, using the wearable device, an opacity of a transparent surface of the wearable device based at least in part on the seizure event prediction data object.
18. The system of claim 16 , wherein the second sub-model is configured to determine the one or more predefined head movement features and the one or more predefined facial movement features based at least in part on a Eulerian video processing technique.
19. The system of claim 15 , wherein the one or more processors are further caused to:
provide an indication of the seizure event prediction data object to a user computing entity; and
store information associated with the seizure event prediction data object to a user profile.
20. The system of claim 16 , wherein the sensor data is captured in real-time by an inertial measurement unit (IMU) of the wearable device.