AI Healthcare Assistant
Practice patient-style conversations and get AI feedback, not a substitute for real clinical training
This Streamlit app is for practicing healthcare communication in a controlled setting. You pick a scenario (such as an initial assessment, explaining a procedure, or discharge-style instructions), respond by text or recorded voice (with transcription), then receive structured feedback the project ties to dimensions like accuracy, clarity, empathy, and completeness. It uses Google Generative AI (Gemini) under the hood.
When it is useful
You want a low-risk rehearsal space for wording and tone, you are studying how model-generated critiques behave, or you are demonstrating AI-assisted coaching in an educational context. It is not a patient record system, telehealth product, or licensure prep guarantee.
What you can do
- Work through documented scenario types and compare text vs. voice practice paths.
- Run analysis after each attempt and review scores and suggestions the app surfaces (see the repository for the exact metric definitions).
- Keep a history of interactions inside the app for your own review, as described in the README.
- Install and run locally following the GitHub repository (API keys and setup are documented there).
Limits
- Not medical advice. The app does not diagnose, treat, or replace licensed supervision; AI output can be wrong, biased, or inappropriate for real patient care.
- Simulation only: scenarios and feedback are generative; they may miss nuance, cultural context, or institutional policy you must follow in practice.
- Regulatory and ethical use (privacy, consent, PHI, local rules) are your responsibility; do not feed real patient identifiers into experimental setups without proper controls.
