Inspiration
Cancer affects millions, and the gap between patients and practical support is still wide. I was inspired to create CancerGPT after witnessing how overwhelming diagnosis and treatment journeys can be. This project is dedicated to helping patients and families feel informed, supported, and less alone.
What it does
CancerGPT is an AI-powered chatbot designed to provide emotional support, answer basic cancer-related queries, guide users to reliable resources, and help them track their treatment journey. It also includes reminders for medication, motivational messages, and access to local resources or helplines.
How we built it
We used:
OpenAI GPT API for conversational intelligence
ReactJS for the frontend UI
Node.js with Express for the backend
MongoDB for storing user conversations (with privacy in mind)
Twilio/WhatsApp integration for social media deployment
The system is trained with curated datasets, FAQs from cancer organizations, and uses prompt engineering to maintain empathetic, informative responses.
Challenges we ran into
Ensuring accuracy without providing medical advice
Filtering misinformation or panic-inducing queries
Balancing empathy with functionality
Designing a minimal yet comforting UI for diverse users
Integrating multilingual support
Accomplishments that we're proud of
Successfully deployed a functioning prototype of CancerGPT
User-tested with 10+ people and received positive feedback
Developed a custom response filter to prevent inappropriate suggestions
Built a feedback loop for model improvement
What we learned
Building healthcare-oriented tools requires extreme care in tone and accuracy
Empathy in AI matters — language choice impacts users deeply
Privacy and trust are as important as technical performance
Deployment on accessible platforms (like WhatsApp) increases impact
What's next for CANCERGPT
Partnering with NGOs and cancer support groups for real-world testing
Adding voice-based interaction for elderly users
Expanding to regional Indian languages
Incorporating symptom journaling and doctor Q&A features
Continuous training based on user feedback and expert reviews
Log in or sign up for Devpost to join the conversation.