Graph-Evolving Meta-Learning for Low-Resource Medical Dialogue Generation

Medical dialogue systems may converse with patients and make a diagnosis automatically. Conventional dialogue generation models cannot be directly applied to such scenarios because of the lack of medical knowledge. A recent study addresses the lack of suitable models in this domain and introduces an end-to-end dialogue system for the medical dialogue generation.

Image credit: pxfuel.com, CC0 Public Domain

Firstly, the conversation is encoded into hierarchical representations. A meta-knowledge graph reasoning network characterizes the correlations among diseases and symptoms, which evolve with new context information. Finally, a response to the requested symptoms is generated.

Moreover, a novel graph-evolving meta-learning framework enables adaptation to new diseases that often have limited data resources. A dataset covering 15 diseases was constructed for the study. Experiments show the superiority of the suggested system over state-of-the-arts.

Human doctors with well-structured medical knowledge can diagnose a disease merely via a few conversations with patients about symptoms. In contrast, existing knowledge-grounded dialogue systems often require a large number of dialogue instances to learn as they fail to capture the correlations between different diseases and neglect the diagnostic experience shared among them. To address this issue, we propose a more natural and practical paradigm, i.e., low-resource medical dialogue generation, which can transfer the diagnostic experience from source diseases to target ones with a handful of data for adaptation. It is capitalized on a commonsense knowledge graph to characterize the prior disease-symptom relations. Besides, we develop a Graph-Evolving Meta-Learning (GEML) framework that learns to evolve the commonsense graph for reasoning disease-symptom correlations in a new disease, which effectively alleviates the needs of a large number of dialogues. More importantly, by dynamically evolving disease-symptom graphs, GEML also well addresses the real-world challenges that the disease-symptom correlations of each disease may vary or evolve along with more diagnostic cases. Extensive experiment results on the CMDD dataset and our newly-collected Chunyu dataset testify the superiority of our approach over state-of-the-art approaches. Besides, our GEML can generate an enriched dialogue-sensitive knowledge graph in an online manner, which could benefit other tasks grounded on knowledge graph.