A professional AI-powered medical intake system that conducts automated pre-visit screening calls and generates comprehensive medical reports. Built with LiveKit Agents, Deepgram STT/TTS, and Google Gemini LLM.
- Structured 4-Section Reports: Primary concern, HPI, medical history, medications
- Maximum 20 Questions: Efficient, focused medical interviews
- OPQRST Framework: Systematic symptom assessment (Onset, Provocation, Quality, Radiation, Severity, Time)
- Medical Terminology: Professional healthcare language
- Pertinent Negatives: Captures what patients DON'T have
- Real-time Speech Recognition: Deepgram STT with Nova-3 model
- Natural Voice Generation: Deepgram TTS with Aura-Asteria model
- Medical Reasoning: Google Gemini 1.5 Flash for clinical logic
- Voice Activity Detection: Optimized Silero VAD for minimal delays
- Background Noise Cancellation: Professional telephony-grade audio processing
- Clean TXT Format: No markdown, healthcare provider friendly
- Structured Data: Organized by medical categories
- Call Details: Complete call metadata and timestamps
- Patient Summaries: Concise information for quick review
- EHR Ready: Easy integration with electronic health records
- Python 3.8+
- LiveKit account and credentials
- Deepgram API key
- Google Gemini API key
- SIP trunk for outbound calling (optional)
- Clone the repository
git clone https://github.com/AliMusawir/Outbound_Caller-.git cd Outbound_Caller-