Developing an intelligent agent for integrated traditional Chinese and western medicine diagnosis and treatment based on clinical practice guidelines is a key approach to advancing the standardization and intelligent deve-lopment of traditional Chinese medicine （TCM） and to supporting clinical decision-making. This paper argues for the necessity of building a knowledge base using guidelines as the sole data source， systematically identifies five core methodological challenges across the entire process， and proposes corresponding solutions. An ontology framework capable of distinguishing between TCM and western medicine concepts should be designed to address the difficulty of knowledge integration. A closed-loop "algorithm-based extraction-expert review" model is adopted to reduce data extraction errors. Expert consensus is incorporated to address decision gaps in specific clinical scenarios outlined in the guidelines， while a multi-stage， standardized output process for the AI system is established to eliminate model hallucinations. Furthermore， an evaluation framework reflecting clinical applicability is developed to enhance the AI system's accuracy and stability. Throughout the research process， it is essential to ensure the in-depth and continuous participation of multidisciplinary experts， strictly control the selection and quality evaluation of guidelines， and rely on the expert consensus method to address decision gaps in the knowledge base. Additionally， continuous validation and iterative optimization of intelligent agents are required to ensure the accuracy and stability of generated outputs. Limitations remain in the mechanisms for real-time knowledge synchronization in intelligent diagnostic and treatment systems， and further validation through large-scale real-world studies is warranted.

[Objective]To clarify the application value of nutritional scoring in patients with sepsis and explore the impact of the modified Nutritional Risk in Critically Ill(mNUTRIC)score on the in-hospital fatality of sepsis patients in the emergency intensive care unit(EICU).[Methods]A retrospective analysis was conducted on the clinical data and laboratory examination results of 436 sepsis patients treated in the EICU of Sun Yat-sen Memorial Hospital of Sun Yat-sen University from January 2021 to May 2024.The patients were divided into survival group(298 cases)and death group(138 cases)according to whether they died or not during hospital treatment,and then compared the two groups'data.Logistic multi-factor regression analysis was used to analyze the risk factors for in-hospital death and the ROC curve to evaluate the predictive value of each risk factor for the prognosis of sepsis patients.[Results]The death group exhibited higher Acute Physiology and Chronic Health EvaluationⅡ(APACHE Ⅱ)score,Nutritional Risk Screening 2002(NRS 2002)score,mNUTRIC score,Sequential Organ Failure Assessment(SOFA)score,Padua Prediction Score for Venous Thromboembolism,the proportions of chronic kidney failure and pneumonia patients,C-reactive protein(CRP)level,lactate concentration and neutrophil count,but lower prognostic nutritional index,cholinesterase level,cholinesterase-albumin ratio and lymphocyte count than the survival group,with statistical significance(all P<0.05).Logistic regression analysis revealed that the mNUTRIC score[OR=1.254,95%CI(1.109,1.417)],CRP[OR=1.004,95%CI(1,1.007)],and pneumonia[OR=1.82,95%CI(1.017,3.257)]were independent risk factors for in-hospital death in sepsis patients.ROC curve analysis showed that the area under the curve(AUC)of the mNUTRIC score for predicting in-hospital death in sepsis patients was 0.683[95%CI(0.623,0.742)],with a sensitivity of 83%and a specificity of 49.3%.The AUC for CRP and pneumonia were 0.602[95%CI(0.533,0.671)],and 0.582[95%CI(0.516,0.647)].[Conclusion]The mNUTRIC score is an independent predictive indicator for in-hospital death in sepsis patients in the EICU.

Objective To explore the effectiveness of FOCUS-PDCA(find,organise,clarify,understand,select,plan,do,check and act)mode in prevention and treatment of incontinence-associated dermatitis(IAD)in emergency intensive care unit(EICU),and to reduce the incidence of IAD.Methods A pre-post control study design was employed in this study.Inpatients admitted to the EICU of the hospital between 1st January and 31st December 2022 were enrolled as study subjects.A total of 169 patients in EICU between January and June 2022 were assigned as a control group,while 197 patients in EICU between July and December 2022 as an trial group.Patients in the control group received conventional quality management,while those in the trial group were treated with the FOCUS-PDCA mode.The two groups were compared in terms of incidence of IAD,the knowledge,trust and behaviour scores of the nurses,and satisfaction from the communications with nurses before and after implementation of the FOCUS-PDCA mode.Results All patients in the two groups had completed the study.Following the implementation of FOCUS-PDCA mode,patients in the trial group exhibited a significantly decrease in incidence of IAD from 10.06%(trial group)to 1.52%(control group),significant increase of scores on the IAD knowledge,behaviour and satisfaction from communications with the nurses,from(36.07±3.98),(16.56±2.15)and(25.15±5.21),to(38.59±3.72),(18.07±1.66),and(30.67±5.84),respectively(all P<0.05).Conclusion FOCUS-PDCA mode can effectively decrease the incidence of IAD,enhance the IAD knowledge and behaviour of nurses and gain satisfactions from communication with nursing staff and clinical quality,thereby improving therapeutic outcomes.