MediAgent

home page if we upload any medical image its going to give ai analysis
this is for vitals predection
this will generates an analysis by body language
this sketch2ai will give body parts pain analysis

Inspiration

During emergency situations—especially in rural areas or with elderly patients—critical symptoms often go unnoticed or misinterpreted. We were deeply moved by the stories of patients unable to express themselves clearly due to pain, unconsciousness, or disability. This inspired us to ask: What if AI could observe, interpret, and assist in understanding body language and symptoms before it's too late?

That’s how MediAgent AI was born—a platform that combines AI, body mapping, and clinical heuristics to help people understand the meaning of symptoms and provide emergency insights in real time.

What it does

Sketch2AI (Body Part Selection) : You tap or click on a body part (like head, chest, leg), and describe the pain. The system tells you the possible cause, intensity, and what to do next. Body Language Translator : You observe someone’s behavior (e.g., sweating, clutching chest, confusion), and the AI tells you what might be happening and gives emergency advice. Symptom Analysis : It combines your input with clinical logic and natural language understanding to give accurate, human-readable explanations. Emergency Recommendations It provides suggestions like "Call emergency", "Monitor for X minutes", or "Possible signs of stroke/heart attack".

How we built it

Frontend ; Used HTML5, Bootstrap 5, and TailwindCSS to create clean, responsive, and professional UI components. Designed button-based inputs for selecting age, consciousness level, body parts, and symptoms—ideal for fast and intuitive use, especially in emergency scenarios. Custom CSS variables were added to maintain a visually consistent theme across the app. Backend : Built using Flask, a lightweight web framework in Python. Managed routes like /sketch2ai and /bodylanguage to process form data and manage state. Used Jinja2 templating to dynamically render the form and results based on user input. Integrated OpenAI's GPT model to analyze symptoms and generate meaningful health interpretations and questions.

Challenges we ran into

We struggled to get consistent and medically relevant answers from the AI initially. Crafting prompts that were: specific enough to guide the AI,yet flexible enough to handle a wide range of inputs was a balancing act that took several iterations Designing a user interface that is: clean, distraction-free, and usable in a high-stress, low-attention environment meant minimal design with maximum clarity. Making buttons intuitive and mobile-friendly took extra UI/UX effort.Flask’s statelessness required us to use hidden form fields and careful route handling to remember selected body parts and pain descriptions without breaking the flow

Accomplishments that we're proud of

Developed an End-to-End Emergency Health Assistant We built a full-stack AI solution that interprets body language and pain descriptions to offer real-time emergency medical advice — all in one unified platform. Our app can take vague descriptions like “he’s dizzy and clutching his chest” and provide concrete insights like “possible heart attack – call emergency immediatel By aligning AI outputs with medically-relevant terminology and severity assessments, we ensured the advice is not just generic but practically useful.

What we learned

The Power of Simplicity in Emergencies We realized that during critical health events, users need immediate, simple, and reliable guidance. This shaped our design decisions — focusing on clarity, minimalism, and responsiveness. We learned that beyond code and models, designing for health means thinking about real people in distress — and building with empathy first.

What's next for MediAgent

Integration with Real-Time Health Devices We aim to connect MediAgent with IoT-enabled health monitors (like heart rate or BP sensors) to offer more accurate emergency triage and dynamic recommendations. Multilingual and Voice-Based Support Many people, especially in rural or low-literacy areas, can’t type detailed symptoms. We're building support for voice input, emotion detection, and local languages to make MediAgent universally accessible. Doctor Collaboration and Validation We’re working with healthcare professionals to validate our logic and suggestions, aiming for regulatory approval as a certified triage assistant in low-resource settings.