JMIR AI · 13 Nov 2026
A Knowledge Graph–Enhanced Question-Answering System for Chinese Patent Medicines in Pediatric Influenza: Development and Evaluation Study
Haoyuan Lih-index 0, Guozhen Zhao, Xingyu Zong, Ning Liang, Huizhen Li, Wandi Zhang, Jiahao Lin, Bohan Ma, Jinghan Jia, Yanping Wang, Nannan Shih-index 4
Citations by database
Each database indexes a different set of journals, preprints and books, so they are shown side by side rather than blended.
- OpenCitations0
- Crossref0
- Europe PMC0
Citations per year
The current year is still counting.
- 2026 so far0
Abstract
Abstract Background Pediatric influenza is an important public health concern. Chinese patent medicines are used in its treatment, but relevant clinical evidence is dispersed across guidelines, expert consensus statements, and clinical studies. The multidimensional knowledge required for syndrome-based prescribing complicates information retrieval and standardized clinical use. Objective This study aimed to develop a traceable knowledge graph of Chinese patent medicines for pediatric influenza, integrate it with a large language model (LLM) to support evidence-based question answering, and evaluate answer quality and preliminary usability. Methods We systematically searched for clinical practice guidelines, expert consensus statements, and clinical studies. Evidence segmentation, ontology development, LLM-assisted extraction with manual verification, terminology normalization, and knowledge integration were used to construct a Neo4j knowledge graph. Structured graph retrieval was combined with retrieval-augmented generation to develop a question-answering system. A standalone LLM and the knowledge graph–enhanced LLM (KG-LLM) each answered 80 standardized questions. Three clinical experts independently evaluated anonymized, randomly ordered answers for core task accuracy, completeness, and unsupported factual content. Five physicians participated in an exploratory usability evaluation using the System Usability Scale (SUS), study-specific Likert items, and self-reported evidence-location outcomes. Results We included 142 publications and generated 1294 evidence units. The knowledge graph contained 3211 nodes and 19,671 relationships, linking structured knowledge to evidence units and source publications. Across 80 paired questions, KG-LLM achieved higher accuracy scores than the standalone LLM (mean 3.05, SD 1.88 vs mean 2.02, SD 1.54; Holm-adjusted P <.001) and higher completeness scores (mean 3.11, SD 1.84 vs mean 2.42, SD 1.62; adjusted P =.002). Among 78 pairs with determinate assessments, unsupported factual content was identified in 11 standalone LLM answers (14.1%) and no KG-LLM answers (0/78, 0%; adjusted P =.002). The mean SUS score was 96.50 (SD 2.85). Of 5 physicians, 4 (80%) reported locating both the original publication and the corresponding evidence. Conclusions The traceable knowledge graph and graph-enhanced question-answering system integrated clinical evidence on Chinese patent medicines for pediatric influenza. KG-LLM achieved higher accuracy and completeness and less unsupported factual content on standardized tasks than the standalone LLM. Preliminary usability findings were favorable, but generalizability, long-term usability, and safety in clinical practice require further evaluation.
What the data shows
- No database records a citation yet, as is usual for a paper this new.
- Nannan Shi has the highest h-index among the authors (4).
Graph
This paper in the middle, its authors and their most-cited papers below, the papers it cites to the left and the papers citing it to the right. The timeline lays every paper out by year.
Select a node for its details. Drag to pan, scroll to zoom, drag a node to move it.
Start with
Cited by 0 papers
Who cites this paper, from open citation data, with Crossref's record of each; the 100 newest when there are more.
References (22)
What it cites, from open citation data, with Crossref's record of each.
Sign in to see 2 more references
A free account opens the rest of the citing papers and references on every paper’s page.