CMKL: Multimodal Continual Learning for Evolving Biomedical Knowledge Graphs

Biomedical knowledge graphs (KGs) are increasingly large, dynamic, and multimodal, driven by rapid advances in biotechnology such as high-throughput sequencing. Machine learning models can infer previously unobserved biomedical relationships and characterize biomedical entities in these graphs, but existing knowledge graph embedding (KGE) methods and their continual learning (CL) extensions either assume static graph structure or fail to exploit multimodal information under evolving data distributions. They also apply uniform regularization across all model parameters, ignoring that different modalities may exhibit distinct forgetting dynamics as the graph evolves. We propose the Continual Multimodal Knowledge Graph Learner (CMKL), a CL framework for biomedical KGs that natively encodes structure (R-GCN), text (frozen BiomedBERT), and molecules (Morgan fingerprints), fuses them through a Mixture-of-Experts (MoE) router, and protects previously learned knowledge with standard EWC regularization and a K-means-diverse multimodal replay buffer. We evaluate CMKL on a 129K-entity biomedical continual benchmark. On continual biomedical entity classification, CMKL reaches Macro-F1 $0.591$ and beats feature-matched continual baselines ($0.414$) by $+43\%$ ($p<10^{-6}$)—an architectural gain isolated from feature availability—preserved across the sequence with near-zero forgetting (AF $0.008$). On continual relationship prediction (real $t_0\to t_1$ evolution tasks), CMKL reaches AP $0.103$, on par with Naive Sequential and EWC ($0.096$) and above Joint Training ($0.078$); a residual-skip fusion variant reaches $0.117$, recovering the structural-only performance ($0.118$) that embedding fusion otherwise sacrifices—fusion that does no harm. Finally, we correct a natural misreading of these graphs: a text-only ablation's high AP ($0.226$) is fully reproduced by a zero-text transductive node-embedding lookup, so the apparent frozen-text “ceiling” is a node-embedding artifact—frozen semantic modalities carry strong classification but negligible link-prediction signal, which the MoE router reflects by down-weighting text on link prediction. Our code is available at https://anonymous.4open.science/r/cmkl-multimodal-ckg-FF56.