ReinCare

Inspiration

Tinder swipes and assembly.ai

What it does

ReinCare is an AI-powered diagnostic and aftercare assistant that bridges the gap between clinical visits and patient recovery.

• For Nurses: It uses real-time audio processing to transcribe consultations and instantly generates structured EMR reports. The nurses can swipe on the patient's symptom questions to help diagnose, and the AI provides a summary based on the answers.
• The Intelligence: ReinCare analyzes these swipes in real-time. If a patient’s swipe indicates a complication, the AI automatically generates a summary and alerts the clinical team.

How we built it

• The Brain: We used Featherless.ai to host DeepSeek, leveraging its reasoning capabilities to transform EMR data into patient-friendly questions and summaries.

• Voice: AssemblyAI handled the heavy lifting of transcribing medical dialogues and speech-to-text clinician tasks in real time.

The Stack: The backend is built with FastAPI (Python)

The UI: Designed for the web using Typescript and React.

Challenges we ran into

• Managing the system architecture
• Adding too many features
• Conflict of ideas

Accomplishments that we're proud of

• Learning Featherless.ai
• Learning AssemblyAi
• Getting the AI backend to work

What we learned

• How to use Featherless.ai to host a model and prompt it.
• How to use AssemblyAI to transcribe speech-to-text in real time.

What's next for blink

• Implementing the patient side

Built With

• css
• flask
• html
• python
• react.js
• supabase
• typescript