The use of traditional Chinese medicine (TCM) has steadily increased every year. In China, TCM accounts for 40% of its pharmaceutical market [1] but unlike Western medicine, TCM herbal medicine is usually administered in the form of a decoction of more than one herbal plant, where each constituent of the plant works synergistically. This inherently produces multiple effects such as antioxidant and anti-inflammatory. While this strategy exemplifies the more accepted polypharmacology paradigm [2], these herbal medicines also inevitably produce several side effects, although it is possible that the combination of herbs may neutralise or counteract the side effects. Unlike Western medicine, TCM herbal medicines are not subjected to intense scrutiny on their efficacy and safety as it is still considered Complementary Medicine. Every year, it is estimated that 230,000 TCM-related adverse effects were reported [3]. Furthermore, herbal products were ranked second as the source of drug-induced liver injury in the United States [4]. Although significant efforts have been made to address these, the safety profile of TCM herbal medicine is still lacking despite the steady increase in its use.

One approach that can be considered in establishing the safety profile of TCM herbal medicine is the use of in silico methods. This is made possible by the availability of public databases that store compound-target/compound-toxicity information that can be used in both mechanisms of action and side effect analyses. Additionally, effort in collecting and storing information on the constituents of the herbal plants and their chemical structure in public databases has expedited the process of analysing traditional medicine compounds. Examples of some of these databases can be seen in Table 1, although it should be noted that this list is not extensive.

In the next sections, the utilisation of these databases in silico to predict and understand the toxicity of TCM herbal medicine will be reviewed. There are three different methods covered in this mini review used for in silico toxicity studies of TCM herbal medicines, which are: 1) machine learning, 2) network toxicology, and 3) molecular docking. It should be noted that this review only covers the intrinsic toxicity of the compounds but not their extrinsic toxicity, where cultivation, processing, storage and distribution may contaminate the herbal raw material.