基于糖尿病足多阶段病程管理的AI智能体构建与验证

打开文本图片集

【Abstract】ObjectiveToinvestigatetheapplicationvalueofatool-drivenartificialinteligence（A）agentinclinical decision-making across multiple stagesof the diseasecoursein patients with diabeticfot.Methods Usingthelargelanguage model（LLM）Qwen3-maxasthereasoningengine，anAIagentadaptabletorepeatedfolow-upscenariosindiabeticfootwas constructedbyintegratingtheReActframework，retrieval-augmentedgeneration（RAG）technologyandmultimodaldataprocesing tools.Validationwasperformedbasedonretrospectivedatafrom34patientswithdiabeticfot（atotalof140visits）. Performancediferences werecomparedamongtheAIagent，thenativelargelanguagemodelQwen3-max，andthemedicalLLM Baichuan-M1-14B.Results The clinical practicality score of the AI agent reached 8.29±0.91 ，significantly higher than that of theQwen3-max（ 7.56±0.70 0 t=4.19 ， P<0.001 ）andBaichuan-M1-14B（ 7.82±0.67 。（剩余17419字）