Objective: To develop QingNangTCM, a specialized large language model (LLM) tailored for expert-level traditional Chinese medicine (TCM) question-answering and clinical reasoning, addressing the scarcity of domain-specific corpora and specialized alignment. Methods: We constructed QnTCM_Dataset, a corpus of 100 000 entries, by integrating data from ShenNong_TCM_Dataset and SymMap v2.0, and synthesizing additional samples via retrieval-augmented generation (RAG) and persona-driven generation. The dataset comprehensively covers diagnostic inquiries, prescriptions, and herbal knowledge. Utilizing P-Tuning v2, we fine-tuned the GLM-4-9B-Chat backbone to develop QingNangTCM. A multi-dimensional evaluation framework, assessing accuracy, coverage, consistency, safety, professionalism, and fluency, was established using metrics such as bilingual evaluation understudy (BLEU), recall-oriented understudy for gisting evaluation (ROUGE), metric for evaluation of translation with explicit ordering (METEOR), and LLM-as-a-Judge with expert review. Qualitative analysis was conducted across four simulated clinical scenarios: symptom analysis, disease treatment, herb inquiry, and failure cases. Baseline models included GLM-4-9B-Chat, DeepSeek-V2, HuatuoGPT-II (7B), and GLM-4-9B-Chat (freeze-tuning). Results: QingNangTCM achieved the highest scores in BLEU-1/2/3/4 (0.425/0.298/0.137/0.064), ROUGE-1/2 (0.368/0.157), and METEOR (0.218), demonstrating a balanced and superior normalized performance profile of 0.900 across the dimensions of accuracy, coverage, and consistency. Although its ROUGE-L score (0.299) was lower than that of HuatuoGPT-II (7B) (0.351), it significantly outperformed domain-specific models in expert-validated win rates for professionalism (86%) and safety (73%). Qualitative analysis confirmed that the model strictly adheres to the “symptom-syndrome-pathogenesis-treatment” reasoning chain, though occasional misclassifications and hallucinations persisted when dealing with rare medicinal materials and uncommon syndromes. Conclusion Combining domain-specific corpus construction with parameter-efficient prompt tuning enhances the reasoning behavior and domain adaptation of LLMs for TCM-related tasks. This work provides a technical framework for the digital organization and intelligent utilization of TCM knowledge, with potential value for supporting diagnostic reasoning and medical education.

Objective To explore the feasibility of blood oxygen level-dependent (BOLD) MRI to detect the chronic hepatitis b-induced early kidney injury. Methods Seventeen clinically diagnosed chronic hepatitis b patients with early kidney injury and 10 healthy volunteers were enrolled in this preliminary study. The 17 patients underwent dynamic nuclear renography and then subdivided into stage 1 kidney injury group (n=7) and stage 2 kidney injury group (n=10). All of the enrolled subjects underwent BOLD examination and T2* relaxation rates (R2*) of renal cortex and medulla of split kidney, and the ratio between them (R2*med/cor) were measured separately. One-way analysis of variance (ANOVA) were performed on the control group and chronic hepatitis b patients group (kidney injury stage 1 and stage 2 group) to compare the difference of renal cortical and medullary R2*values and R2*med/cor ratio. ROC curves were used to evaluate the efficacy of renal cortical and medullary R2* values and R2*med/cor ratio to diagnose the chronic hepatitis b-induced kidney injury. Results The cortical R2*values of control group, stage 1 kidney injury group and stage 2 kidney injury group were(16.87 ± 0.74)/s,(17.88 ± 0.73)/s,(20.29 ± 2.87)/s, respectively;the medullar R2*values of control group, stage 1 kidney injury group and stage 2 kidney injury group were (28.07±1.03)/s,(31.14±2.49)/s,(32.81±3.28)/s, respectively;R2*med/cor of the of control group, stage 1 kidney injury group and stage 2 kidney injury group were 1.67 ± 0.09, 1.75 ± 0.16, 1.63 ± 0.13, respectively, and the differences among the three groups were statistically significant (F values were 17.779, 19.170 and 3.439 , all P<0.05). Furthermore, the renal cortical and medullary R2* values of chronic hepatitis b patients were significantly higher than the control group, and the the renal cortical R2* value of the patients in stage 2 kidney injury group was also higher than the stage 1 kidney injury group. The area under curve (AUC) of ROC of the renal cortical and medullary R2*values and R2*med/cor to diagnose chronic HBV hepatitis-induced early kidney injury were 0.903, 0.949 and 0.526, respectively. Conclusion It's feasible and has great value to use renal BOLD MRI for the diagnosis of chronic hepatitis b-induced early kidney injury, and the renal cortex is more sensitive than the medulla to the kidney injury.