CookiesNChopsticks

Inspiration

We were inspired by the challenges faced by the elderly community, especially those living with Alzheimer’s + other disabilities. Many seniors live independently, but medication adherence and safety become major concerns. Caregivers can’t be present 24/7, and missing even one dose can have serious consequences. We wanted to build something that gives families peace of mind while preserving independence.

What it does

PillGuard is an AI-powered smart pill dispenser that not only reminds patients to take their medication but also verifies what happened.

Using a camera and real-time AI vision, PillGuard can detect whether a patient:
Took their medication
Opened the box but didn’t take it
Never showed up
Is in physical distress
It then sends real-time alerts to caregivers and logs everything into a dashboard that tracks behavior over time.

How we built it

signed PillGuard as a full-stack system combining hardware, AI, and web technologies: Hardware (Raspberry Pi):

Ultrasonic sensor detects pill box interaction
Webcam records a short clip
LED, buzzer, and LCD provide real-time patient feedback
AI (Claude Vision API):
Frames are sampled and classified into states like TOOK_PILL, NO_TAKE, DISTRESS, or NO_SHOW
A majority voting system determines the final outcome Backend (FastAPI):
Receives events from the Pi
Logs data into a SQLite database
Sends alerts via SMS (Twilio)
Runs scheduling logic for missed doses Frontend (React Dashboard):
Displays live alerts and medication logs
Shows adherence trends over time
Allows caregivers to manage schedules and settings

Challenges we ran into

Since we started with just an idea, one of the biggest challenges was designing a system that could reliably interpret real-world human behavior by using the Raspberry Pi. It was hard to connect our raspberry pi to our system using claude. We also had a hard time handling hardware inconsistencies (sensor noise, camera angles). We tried reducing latency between detection and alerts## Accomplishments that we're proud of

Designing a system that goes beyond reminders and actually understands patient behavior
Integrating hardware, AI, and a full-stack dashboard into one cohesive solution for an actual problem that affects not only our loved ones personally, but also families just like our own.
Building a concept that not only solves a daily problem but also provides long-term health insights
Creating a product that has both technical depth and real human impact

What we learned

How to architect an end-to-end system combining Raspberry Pi 3, motion sensing camera system, screen to display reminders, sound system, IoT, AI, and web apps
The importance of designing for real users, especially elderly individuals
How small daily interactions can become powerful data signals over time
The value of building technology that is proactive, not just reactive

What's next for CookiesNChopsticks

Next, we plan to:

Build and test a working prototype with real users
Improve AI accuracy with more training data and edge-case handling
Add features like voice interaction and multilingual support
Integrate with healthcare providers for deeper insights
Expand the data analytics to better predict cognitive decline trends
Our long-term vision is to turn PillGuard into a complete AI-assisted home care system that helps families stay connected and proactive—no matter the distance.