MedCognis

Inspiration

from daily rush of queue in hospital during emergency.

What it does

The system addresses a core operational challenge in emergency and inpatient care: accurate, transparent, and real-time patient prioritization.

How we built it

some by vibe coding and all by knowledge of system design and creative thinking.

Challenges we ran into

finding the datasets (real-time) and performing model training and accessing diseases that run undefined and patient may suffer still from that disease (The basic reason is that common problem can be solved but not new disease like accident cases).

Accomplishments that we're proud of

Proud to deliver a prototype that help people to fix their daily rush into the hospital for basic things too.

What we learned

-model training (XG-Boost)

• local LLM (Llama 3 via Ollama ) -Explainable SAI (SHAP)
• dataset valuation and manipulation
• Creative thinking
• fixing bugs in fixed time interval
• collaboration (teamwork)

What's next for MedCognis

• Model training on real datasets
• Performance improvement
• latency reduction
• Explainable AI support (voice agent support)
• Multilanguage AI chatbot support
• fixing temporary bugs
• improving and enhancing the traffic into the website
• testing in real world by developing into cloud.

Built With

• animations:
• css
• data
• fastapi
• framer
• hooks
• ml-model
• next.js
• node.js
• react
• recharts
• styling:
• tailwind
• ui
• uvicorn-ai
• v4
• visualization: