Pandora

Inspiration

The idea for Pandora was born from a deeply personal place. My sister’s first experience with period cramps was a nightmare the kind that ends in the emergency room. Seeing her go through that much pain, confusion, and fear lit a fire in me. It made me realize how little technology exists to truly support people dealing with menstrual pain, and how urgently we need smarter, more compassionate solutions. Pandora is our answer to that need.

What it does

Pandora is an AI-powered TENS (a technology from the 60’s that’s used for pain relief) belt designed to provide personalized relief from period cramps. Here's how it works:

The user starts by chatting with our integrated AI chatbot, which collects detailed information about symptoms, cycle phase, and pain intensity.
This data feeds into a hybrid AI system:
- A base model for general predictions
- A user-specific online learning model that adapts over time
As the system learns more about the user, the online learner takes over predicting and adjusting TENS frequency, pulse width, and heat levels in real time for optimized relief.

No more guessing. No more one-size-fits-all solutions.

How we built it

Pandora is a full-stack solution combining hardware, AI, and user experience design.

Hardware

ESP32 microcontroller for wireless communication and real-time control
TENS unit (in house built*)* for adjustable electrical stimulation
Heating pad with regulated temperature control
Rechargeable battery and ergonomic belt form factor

Software

AI Chatbot : Gathers symptom data conversationally
Hybrid AI Model: Built using sickit learn and River
Flutter Mobile App: For user control, manual overrides, and session tracking
Azure Backend: Cosmos DB, Azure Functions, IoT Hub for real-time data handling and communication

Challenges we ran into

Designing a machine learning model that could adapt in real time with limited data
Ensuring safety in electrical stimulation and heat delivery
Converting subjective user feedback into actionable input for the AI
Keeping latency low between user input, prediction, and physical response

Accomplishments that we're proud of

Second place winner at the ARB (African Robotic Brains) competition
Successfully built and demoed a real-time pain relief system that learns from the user
Bridged AI, healthcare, and embedded systems into one functional and meaningful product
Sparked conversations about the importance of personalized tech in women’s health

What we learned

Building for real people requires empathy first, tech second
Even simple biometric data can unlock powerful personalization with the right AI model
Hardware-software integration demands precision and patience
User trust is key especially in health devices worn on the body

What's next for Pandora

Expanding the AI model to include hormonal patterns and environmental factors
Partnering with health professionals for clinical validation and feedback
Redesigning the hardware to be smaller, more flexible, and discreet
Launching a beta testing program with real users
Exploring integrations with wearables and health tracking apps for smarter cycle management