Knowledge Graph-based JingFang Drug Efficacy Analysis With a Supportive Randomized Controlled Influenza-like Illness Clinical Trial

Abstract

This paper presents a novel methodology for drug efficacy analysis using a knowledge graph, validated by a randomized controlled clinical trial. To provide a comprehensive understanding of drug treatment effects, a learning-based workflow is developed to mine drug-disease entities and relations from literature. These relations build a knowledge graph used for clustering-based drug efficacy analysis. Our tool reports the learned relatedness between drugs and diseases, indicating efficacy levels. JingFang is identified as effective for flu and colds. To validate this, a clinical trial was conducted on Influenza-like illness. Between August 25 and October 12, 2020, 106 patients were randomly assigned in a 1:1 ratio to either the combined group (53) or the control group (53). Patients in the combined group received Xinkangtai Ke and JingFang, while the control group received Xinkangtai Ke only for 7 days. The combined group's cure rate was 92.5% (49) compared to 81.1% (43) in the control group (p=0.0852). The very effective rate was 98.1% (52) in the combined group versus 92.5% (49) in the control group (p=0.3692). For middle-aged and elderly participants, the combined group's recovery rate was significantly higher than the control group's (100% vs 78.4%, p=0.0059, 95% CI: 21.6 (8.3, 38.2)). No adverse effects were observed in either group. The results indicate that JingFang is effective for patients with Influenza-like illnesses, especially those over 34 years old. This study highlights the potential of knowledge graph-based analysis in drug efficacy research.