Article type
Year
Abstract
Background: herb-induced liver injury (HILI) is difficult to evaluate and identify, particularly as herb-related toxicity is often brought about by the unique composition of the medication and the complexity of its clinical use.
Objectives: on the premise of not interfering with any researchers’ work, this study sets out to establish an indirect risk prediction and evidence-based evaluation method for Chinese herbal medicine, with the aim of improving the safety of clinical diagnosis and treatment in traditional Chinese medicine (TCM).
Methods: first, we set up a multi-dimensional data acquisition platform (to determine the evidence for herb-related toxicity, along with research into reactive metabolites (RMs) and individual research evidence of hepatotoxicity in TCM). This platform relied on extraction technology based on RMs. Second, we pooled all information regarding different types of evidence into a single database to standardize the acquisition methods for clinical hepatotoxicity studies, thus improving the overall accuracy. Third, enlightened by the principle and associated technology of the 'neural network algorithm', we could develop an artificial intelligence algorithm, a risk prediction model for individualized clinical hepatotoxicity, enabling machine learning for different items to be established by neural network extrapolation, so that similarities in measurements could be compared.
Results: we combined multiple techniques in a dynamic way and applied them to prospective (basic and clinical) studies, and, during the study, we established a new technology for hepatotoxicity acquisition, analysis and risk prediction.
Conclusions: in this way, it is expected that an effective, scientific and objective evaluation of herb-related hepatotoxicity can be developed, and that potential risk signals for toxic biomarkers can be identified. Thus it should be possible to predict which of these may exhibit individual hepatotoxicity, providing a new way for investigating the early evaluation of herb-related hepatotoxicity in general.
Patient or healthcare consumer involvement: participating patients have read and approved this version of the research report, and due care has been taken to ensure the integrity of the work.
Objectives: on the premise of not interfering with any researchers’ work, this study sets out to establish an indirect risk prediction and evidence-based evaluation method for Chinese herbal medicine, with the aim of improving the safety of clinical diagnosis and treatment in traditional Chinese medicine (TCM).
Methods: first, we set up a multi-dimensional data acquisition platform (to determine the evidence for herb-related toxicity, along with research into reactive metabolites (RMs) and individual research evidence of hepatotoxicity in TCM). This platform relied on extraction technology based on RMs. Second, we pooled all information regarding different types of evidence into a single database to standardize the acquisition methods for clinical hepatotoxicity studies, thus improving the overall accuracy. Third, enlightened by the principle and associated technology of the 'neural network algorithm', we could develop an artificial intelligence algorithm, a risk prediction model for individualized clinical hepatotoxicity, enabling machine learning for different items to be established by neural network extrapolation, so that similarities in measurements could be compared.
Results: we combined multiple techniques in a dynamic way and applied them to prospective (basic and clinical) studies, and, during the study, we established a new technology for hepatotoxicity acquisition, analysis and risk prediction.
Conclusions: in this way, it is expected that an effective, scientific and objective evaluation of herb-related hepatotoxicity can be developed, and that potential risk signals for toxic biomarkers can be identified. Thus it should be possible to predict which of these may exhibit individual hepatotoxicity, providing a new way for investigating the early evaluation of herb-related hepatotoxicity in general.
Patient or healthcare consumer involvement: participating patients have read and approved this version of the research report, and due care has been taken to ensure the integrity of the work.