Introduction/Inspiration
Bumi is a wearable bracelet for children with Tourette's syndrome that alerts parents when a fully incapacitating tic attack is occurring, allowing children to receive medication faster. To preface, Tourette’s syndrome (TS) is a neurological disorder characterized by repetitive and unwanted movements, otherwise known as tics. Tic attacks are periods of non stop tics that can last anywhere from a few minutes to several hours. We learned that communicating during these tics attacks becomes nearly impossible as a Tourette’s patient is incapacitated. However, medication given during a tic attack can decrease effects and shorten the span of a tic attack significantly. Bumi is designed for children in hopes of providing them with care early on in the attack and preventing unintentional self-harm and struggle.
Tools and Technologies
In order to detect tic attacks for Tourette's Syndrome patients, we deploy a custom built machine learning algorithms to detect attacks with precision and accuracy. Our Local Outlier Factor algorithm is custom build for each individual patient. After two days of data gathering and calibration from our three sensors, our algorithm creates a custom prediction mechanism for each patient. This mechanism can be used to find health vitals and movement changes which are clear indicators of tic attacks. The program is automatically executed as it is hosted in google cloud services using the sklearn and seaborn.
Studies have shown that variation in body temperatures, variations in sweat as well as pulse rate occur during tic attacks. We gather data from our three sensors: Gyroscope/Accelerometer, Temperature/Humidity, and Pulse Sensor. This data from the smartband is collected and sent over Wifi to a cloud system through the ESP32 Feather. The cloud system used for the purposes of our product is a Firebase Realtime Server, which collects the following data: Acceleration, Angle Roll, Angle Pitch, Humidity, Pulse and Temperature. The acceleration data is gathered from the accelerometer and accounts for the speed of movement during tic attacks. The Angle roll and Pitch account variation in movement is collected from the Gyroscope sensor. Additionally, we have a button feature on our smartwatch in order to signal alerts if patients need instant assistance and the watch does not pick up an attack. It needs to be pressed and held for around 5 seconds to activate; the delay is intended to prevent false alerts from being sent out.
Our application was created from scratch using Flutter. This companion app is intended for parents and caretakers to monitor a patient/child's health. It broadcasts live data, namely pulse and body temperature, by pulling it through Firebase and updating in real time. It also has an overall rating of motion stability with three levels: stable, slightly unstable, and very unstable. These are levels calculated through the values read from the accelerometer and gyroscope. Once a tic attack is detected, a notification bar at the bottom will change from reading "All Good!" to "Attack Occurring!" that lets the caretaker know something is wrong.
Challenges
The hardest part of our project was in regards to size and scale. Due to limitations placed on hardware components, we were not able to scale our product down to fit it all on a wrist. This led to a larger design than planned. If we had known this limitation earlier, we would have planned on adding a LED screen among other similar features as well.
Next Steps
Clearly, there is a demonstrated need for Bumi among the target market. We have a niche, technologically complex product made to make the lives of children with TS, a long overlooked segment, much easier. Bumi will become the foundation for other wearable tech suited to more neurological disorders, and has the potential to change the world of healthcare. We want to take our product to the next level by making it more sleek and wearable, and eventually reach this goal.
Log in or sign up for Devpost to join the conversation.