USING MACHINE LEARNING TO ALLOW EPILEPTICS TO DRIVE
DIGITAL MEDICINE & WEARABLE TECHNOLOGY
Author: Lauren Spalding
Status: Project Concept
Background: Epilepsy is an increasingly developing disease, with over 65 million people affected. Those with epilepsy are at risk for a seizure at any given moment, therefore they are limited to everyday activities such as driving, especially independently. In each state, the driving regulations differ, however all patients are at risk of experiencing an epileptic attack while driving. If the patient was to have a seizure while driving, they would quickly lose control of the vehicle and potentially cause trauma to themselves and others around them. This puts people with epilepsy in a situation that deprives them of their independence, as they can no longer drive themselves or others.
Goal: The goal of this project idea is to use supervised machine learning to detect the pre-symptoms of an epileptic seizure due to the pattern recognition of each patient’s individual seizure. After detecting that the patient would have a seizure, it would alert the patient to pull over to the side and then notify either an emergency medical technician or the primary person who is caring for that patient. This would provide an accurate detection due to the way that the machine functions, as it can recognize facial and physical changes in the way the patient acts.
Methods: The machine learning tool would first analyze a video taken of the seizures that the patient experiences on a day-to-day basis. Through doing this, the system would be able to detect how a patient’s physical and emotional state changes before the seizure starts. The system would track the patterns and then apply them to when the person is driving. These patterns can include jerking, excessive staring, or facial abnormalities. Due to the individualization aspect of the seizure, the machine would recognize the specific indications that are personal to that single patient. Once the machine detects that the person will be having a seizure, it would alert the patient to pull their vehicle over to the side lane and completely stop the car. Once the computing system recognizes that the patient is actually experiencing the seizure, it would either notify an emergency medical team or the person who is primarily taking care of the patient. This connects back to the individualized part of epilepsy, as patients have different ways of using others to manage their epilepsy. The machine would be programmed to call or notify whatever person/team that the patient needs, as ambulance service has the ability to be expensive, therefore it would be unnecessary for the system to notify EMTs each time, especially if the patient does not need that type of help each time that they have a seizure.