基于入院指标的细菌性肝脓肿液化成熟度贝叶斯统计预测模型构建

背景和目的：脓肿的液化成熟度是影响细菌性肝脓肿（PLA）患者早期治疗、有创引流及预后康复的一个重要因素。在疾病早期能够有效诊断PLA并及时给予相应的评估与治疗是临床诊疗的重点和难点。目前，国内外的诊疗策略是通过增强CT、MRI检查及手术情况确定脓肿性质，缺乏快速确定脓肿性质的手段。本研究利用入院常规检查指标，通过采用贝叶斯统计方法，构建PLA液化成熟度的预测模型，为PLA的早期诊断和治疗提供科学依据。方法：收集2018年1月—2022年12月期间新疆医科大学第五附属医院收治的116例PLA患者资料。根据增强CT、手术情况明确的脓肿成熟度，将患者分完全液化组（59例）与未完全液化组（57例），比较两组患者入院常规检查指标和临床特征。对原始资料进行二分类，经筛选后得出诊断价值较大的指标。采用贝叶斯统计法建立PLA液化成熟度预测模型，并选取2023年1月—2023年11月期间新疆医科大学第五附属医院接诊的23例PLA患者对模型进行验证，并生成受试者工作曲线（ROC）评估本预测模型的预测效能。结果：筛选结果显示，发病时间、白细胞计数、中性粒细胞计数、中性粒细胞百分比、中性粒细胞与淋巴细胞比值、...

Construction of a Bayesian statistical predictive model for the liquefaction degree of pyogenic liver abscess based on admission indexes

Background and Aims The liquefaction degree of abscesses is a crucial factor affecting the early treatment, invasive drainage, and prognosis of patients with pyogenic liver abscesses (PLA). Effectively diagnosing PLA early and providing timely assessment and treatment are focal challenges in clinical practice. Currently, the diagnostic and treatment strategies both at home and abroad rely on enhanced CT scans, MRI examinations, and surgical conditions to determine the nature of abscesses, and there is a lack of rapid means to determine abscess characteristics. This study was conducted to construct a predictive model for the liquefaction maturity of PLA using routine admission examination indexes and the Bayesian statistical method to provide a scientific basis for the early diagnosis and treatment of PLA.Methods Data of 116 PLA patients admitted to the Fifth Affiliated Hospital of Xinjiang Medical University between January 2018 and December 2022 were collected. Patients were classified into a complete liquefied group (59 cases) and an incomplete liquefied group (57 cases) based on the abscess maturity confirmed by enhanced CT and surgical conditions. Comparison was made between the two groups regarding routine admission examination indexes and clinical characteristics. The original data was subjected to binary classification, and after screening, variables with significant diagnostic values were identified. The Bayesian statistical method was employed to establish a predictive model for the liquefaction degree of PLA. The model was validated using 23 PLA patients admitted to the Fifth Affiliated Hospital of Xinjiang Medical University from January 2023 to November 2023, and the ROC curve was generated to evaluate the model''s predictive performance.Results Screening results revealed that factors such as onset time, white blood cell count, neutrophil count, neutrophil percentage, neutrophil-to-lymphocyte ratio, platelet count, procalcitonin, alanine aminotransferase, and plain CT values were significantly associated with the liquefaction degree of PLA (all P<0.05). ROC curve validation demonstrated that the Bayesian statistical predictive model based on these variables had a sensitivity of 90.0%, specificity of 84.6%, and accuracy of 87.3%.Conclusion The constructed Bayesian statistical predictive model for the liquefaction degree of PLA can effectively and rapidly determine the nature of abscesses. It can be used in the early stages of the disease when PLA is not excluded based on routine examination indicators at admission and clinical features with good sensitivity and specificity.