This system integrates Whisper speech recognition and Qwen large models to achieve multimodal intelligent management. Firstly, structured medical records were generated through real-time speech tranion, combined with SNOMED CT standardized terminology. Then, diagnostic and treatment text were automatically generated by means of generative AI, with an RAG architecture for dynamic error correction. Finally, the quality control center integrates timeseries analysis and causal reasoning to automatically generate compliance reports. Its technical features include low-latency speech parsing, multimodal semantic alignment, and privacypreserving computation, running on NVIDIA A800 cloud clusters.

Technology provider:South China University of Technology