Enhancing Biomedical Relation Extraction with Transformer Models using Shortest Dependency Path Features and Triplet Information

•An effective and simple approach to improve pre-trained transformer models for biomedical relation extraction.•Selecting the representative samples using Shortest Dependency Path features and extraction of triplet features in model training.•Handling both intra-sentential and inter-sentential relation extraction.•The results on the Chemical-Induced Disease Relation (CDR) dataset improve the state of the art model performances, specifically in tackling inter-sentential relations.