Collaborating LLMs and PLMs for Medical Tasks

The intelligentization of the medical field not only enhances service quality and efficiency but also reduces costs and meets the growing demand for healthcare. Although various AI models have been developed to address specific medical tasks, they struggle to independently manage complex tasks. Given the strengths of large language models (LLMs) in problem understanding and logical reasoning, we propose MedicalGPT, which leverages fine-tuned LLMs to connect and orchestrate different medical AI models to solve complex medical tasks. Specifically, upon receiving a patient request, MedicalGPT employs LLMs for task planning, selects appropriate models to handle each subtask, and integrates the results to generate the final output provided to the patient, thereby advancing the intelligentization of healthcare.