Access to reliable health consultation is a critical global challenge. While large language models show promise, their practical utility is typically limited by high input barriers, lack of adaptive response strategies for complex scenarios, and insufficient real-world evaluation. To address these gaps, we present a novel health AI assistant that integrates a low-latency full-duplex voice interface with a Multi-Agent System (MAS). The full-duplex design supports natural turn-taking, significantly lowering interaction barriers and quadrupling average conversation depth from 2.62 to 10.16 turns compared to text-based baselines. Furthermore, our MAS architecture decouples clinical reasoning from general conversation, effectively managing complex queries and boosting clinical performance scores by over 30% compared to single-agent baselines. We conducted the first large-scale analysis of an LLM-based health assistant deployed to the general public (N > 5,000), demonstrating superior user engagement and providing empirical evidence on the safety and utility of deploying such systems in realistic environments.

With the development of medical digitization, the extraction and structuring of Electronic Medical Records (EMRs) have become challenging but fundamental tasks. How to accurately and automatically extract structured information from medical dialogues is especially difficult because the information needs to be inferred from complex interactions between the doctor and the patient. To this end, in this paper, we propose a speaker-aware co-attention framework for medical dialogue information extraction. To better utilize the pre-trained language representation model to perceive the semantics of the utterance and the candidate item, we develop a speaker-aware dialogue encoder with multi-task learning, which considers the speaker’s identity into account. To deal with complex interactions between different utterances and the correlations between utterances and candidate items, we propose a co-attention fusion network to aggregate the utterance information. We evaluate our framework on the public medical dialogue extraction datasets to demonstrate the superiority of our method, which can outperform the state-of-the-art methods by a large margin. Codes will be publicly available upon acceptance.