HDI Highlighter, The First Intelligent Tool to Screen the Literature on Herb-Drug Interactions

Herbal food supplements are commonly used and can be an important part of patient self-care. Like all other bio-active and therapeutic products, they have a benefit/risk balance. These products are not without adverse effects and potentially interact with other therapies. Educating patients and providing information for health professionals about the risk of herb-drug interactions is key. One of the purposes of the biomedical literature is to inform prescribers. Scientific literature accessible on databases such as PubMed is dense and careful reading is time consuming. We propose a reading aid tool named "HDI highlighter" to help readers to find key information in clinical studies and case reports describing herb-drug interactions. It uses natural language processing algorithms (artificial intelligence) with a pharmaceutical focus. Semantic relation extraction for herb-drug interactions from the biomedical literature are overexpressed using keywords. We have tested it to review 120 published articles over the last 10 years. In these articles, we have shown that case reports often involved long-term or semi-long-term treatments such as cancer or human immunodeficiency virus therapies, antiepileptic drugs, or central nervous system drugs. Similarly, these classes of drugs are more extensively targeted by clinical studies. Herb-drug interactions described in case reports are identified in medicinal, recreational, and alimentary uses. They also usually lack a rigorous description of the herb(s) involved. Typically, clinical studies provide a complete description of protocols and dosages, with a few exceptions explained by patients' needs. Clinical studies on herbs are nevertheless conducted on a limited number of patients. All these limitations make the interpretation of herb-drug interactions complicated, but the HDI highlighter provides a quick overview of the herb-drug interaction literature.