MedCare: Advancing Medical LLMs through Decoupling Clinical Alignment and Knowledge Aggregation
This work addresses the problem of generalization in medical AI for healthcare applications, representing an incremental improvement by refining existing fine-tuning approaches.
The paper tackles the challenge of medical LLMs struggling with both knowledge-intensive and alignment-required tasks by proposing MedCare, a two-stage fine-tuning pipeline that decouples clinical alignment and knowledge aggregation, achieving state-of-the-art performance on over 20 medical tasks and specific alignment tasks across model sizes.
Large language models (LLMs) have shown substantial progress in natural language understanding and generation, proving valuable especially in the medical field. Despite advancements, challenges persist due to the complexity and diversity inherent in medical tasks, which can be categorized as knowledge-intensive tasks and alignment-required tasks. Previous approaches either ignore the latter task or focus on a minority of tasks and hence lose generalization. To address these drawbacks, we propose a progressive fine-tuning pipeline. This pipeline employs a Knowledge Aggregator and a Noise aggregator to encode diverse knowledge in the first stage and filter out detrimental information. In the second stage, we drop the Noise Aggregator to avoid the interference of suboptimal representation and leverage an additional alignment module optimized towards an orthogonal direction to the knowledge space to mitigate knowledge forgetting. Based on this two-stage paradigm, we proposed a Medical LLM through decoupling Clinical Alignment and Knowledge Aggregation (MedCare), which is designed to achieve state-of-the-art (SOTA) performance on over 20 medical tasks, as well as SOTA results on specific medical alignment tasks. Various model sizes of MedCare (1.8B, 7B, 14B) all demonstrate significant improvements over existing models with similar model sizes.