BioBridge: Bridging Biomedical Foundation Models via Knowledge Graphs
This addresses the problem of isolated unimodal models in biomedical AI, enabling better integration for tasks like drug discovery and question answering, though it is incremental as it builds on existing foundation models.
The paper tackles the limitation of unimodal biomedical foundation models by introducing BioBridge, a parameter-efficient framework that bridges these models using knowledge graphs to enable multimodal behavior, achieving an average 76.3% improvement over baseline methods in cross-modal retrieval tasks.
Foundation models (FMs) are able to leverage large volumes of unlabeled data to demonstrate superior performance across a wide range of tasks. However, FMs developed for biomedical domains have largely remained unimodal, i.e., independently trained and used for tasks on protein sequences alone, small molecule structures alone, or clinical data alone. To overcome this limitation of biomedical FMs, we present BioBridge, a novel parameter-efficient learning framework, to bridge independently trained unimodal FMs to establish multimodal behavior. BioBridge achieves it by utilizing Knowledge Graphs (KG) to learn transformations between one unimodal FM and another without fine-tuning any underlying unimodal FMs. Our empirical results demonstrate that BioBridge can beat the best baseline KG embedding methods (on average by around 76.3%) in cross-modal retrieval tasks. We also identify BioBridge demonstrates out-of-domain generalization ability by extrapolating to unseen modalities or relations. Additionally, we also show that BioBridge presents itself as a general purpose retriever that can aid biomedical multimodal question answering as well as enhance the guided generation of novel drugs.