About the Project: Insight Onco Inspiration Cancer is among the major causes of mortality globally, but early detection is still a problem, particularly in rural areas and for those who cannot afford high-cost healthcare. Our group, The Synapse, aimed to create an affordable, AI-based cancer prediction platform that would empower individuals with early diagnosis and actionable insights, eventually boosting survival rates.
How We Built It Design: We initially designed our UI/UX in Figma for a simple and intuitive interface. Development: We developed our platform on Next.js for smooth performance and hosted it on Vercel for efficient and scalable hosting. Machine Learning Models: We trained four models for detecting cancer: Skin Cancer: 92% accuracy (Image-based) Oral Cancer: 89% accuracy (Image-based) Lung Cancer: 94% accuracy (Questionnaire-based) Cervical Cancer: 96% accuracy (Questionnaire-based) We used publicly available research-supported datasets for training the models. FHIR Integration: We intended to make our platform SMART on FHIR-compatible so that Electronic Health Records (EHR) are stored securely with GDPR & HIPAA compliance. Data Handling: We securely store and update user predictions on the FHIR server using MeldRx, keeping it accessible to healthcare providers.
What We Learned The significance of healthcare data security and HIPAA & GDPR standard compliance. Improving AI/ML models for practical healthcare uses. Enabling SMART on FHIR for improved interoperability with healthcare records. Designing a scalable, easy-to-use platform for non-technical consumers.
Challenges We Overcame Accuracy & Speed Tradeoff: Refining our AI models to make rapid but extremely accurate predictions. Complexity of FHIR Integration: Establishing secure, standardized data storage in the healthcare environment. Making It User-Accessible: Building a clean, mobile-optimized experience for technically less savvy users.
The Impact With Insight Onco, we are embarking on democratizing early cancer detection so that no one is left behind based on geographical or economic reasons. We aim to make life-saving cancer predictions accessible, precise, and secure.
The adventure has just started, and we're thrilled to advance the frontiers of AI in healthcare!
Built With
- canva
- figma
- kaggle
- machine-learning
- mongodb
- node.js
- python
- scikit-learn
Log in or sign up for Devpost to join the conversation.