Inspiration
Having co-workers, clients and friends with dementia, the team has witnessed first hand the difficulties of living with a neurodegenerative disorder. We wanted our hack to help improve the daily life of these friends.
What it does
When activated, the system uses the acceleration of the door to detect an adult with dementia leaving the house unattended, and alert their family to the situation. This solution helps manage the situation while letting users maintain their autonomy.
How we built it
- the embedded system detects the motion of the house door opening using an IMU and Arduino UNO. It sends the acceleration data of the IMU to the Serial Monitor
- the Python script accesses the acceleration data through and Arduino-python bridge and compares it to training data to classify it as a significant door opening (this is done using our developed pattern recognition algorithm)
- the Python script sends caregivers a text message alerting them to the situation using Standard Library. It also sends the older adult a text message linking them to a website that can guide them in case they need assistance
- the website (built using HTML, CSS and JS) shows the user their current location using Google Maps API from Google Cloud Services to gently guide them home, while offering the option to contact EMS and family members through a simple button press
Challenges we ran into
- finding useful features when developing pattern recognition algorithm
- soldering the IMU to the breadboard
- integrating scripts from four languages into a fully automated system
Accomplishments that we're proud of
- 95% accuracy rate in detecting door opening
- programming pattern recognition algorithm from scratch
- Beautiful user-oriented design for web app
- ability to help keep our friends with dementia safe, while retaining their autonomy
What we learned
The details in integrating and designing take time! Testing system for edge cases is crucial to successful deployment.
What's next for Lighthouse
Real user testing (on a real door) to provide insightful feedback in developing a higher-fidelity prototype
Log in or sign up for Devpost to join the conversation.