Jinghao Lin

EasyLens amplifies subtle lesion representations in medical VLMs without additional training, dramatically improving detection rates in challenging clinical scenarios.

Fragmented medical data hurts MLLM performance: this paper shows how a hierarchical medical knowledge graph can be used to engineer training data that substantially improves MLLM accuracy on complex clinical tasks.

MedXIAOHE leapfrogs closed-source systems on medical benchmarks by using entity-aware pretraining and RL-based reasoning, offering a new open-source foundation for medical AI.