目的分析急性A型主动脉夹层(acute type A aortic dissection,ATAAD)术后早期急性胃肠道损伤(acute gastrointestinal injury,AGI)的危险因素及其预后,并构建AGI发生风险列线图疾病预测模型。方法纳入2016—2021年在兰州大学第一医院行A...目的分析急性A型主动脉夹层(acute type A aortic dissection,ATAAD)术后早期急性胃肠道损伤(acute gastrointestinal injury,AGI)的危险因素及其预后,并构建AGI发生风险列线图疾病预测模型。方法纳入2016—2021年在兰州大学第一医院行ATAAD体外循环手术患者,并将患者分为AGI组和non-AGI组。比较两组患者的临床资料。使用R语言构建列线图并验证模型的预测能力。结果共纳入188例患者,其中男166例、女22例,年龄22~70(49.70±9.96)岁。AGI组60例,non-AGI组128例。通过多因素logistic回归分析,两组患者的主动脉夹层(aortic dissection,AD)危险评分、肠系膜上动脉(superior mesenteric artery,SMA)灌注不良、主动脉阻断时间及术中输注红细胞均为AGI的预测因素(P<0.05)。两组在术后呼吸机辅助时间、住ICU时间、肝功能不全、肾功能不全、肠外营养、院内感染以及术后30 d死亡方面差异均有统计学意义(P<0.05)。纳入经多因素logistic回归分析筛选的预测因素建立列线图,C指数为0.888,通过内部验证C指数为0.848。采用受试者工作特征曲线评估模型区分度,曲线下面积为0.888。结论ATAAD术后AD危险评分、SMA灌注不良、主动脉阻断时间及术中输注红细胞均为AGI的预测因素,建立的预测模型具有良好的预测能力。展开更多
合并症预测是典型的多标签分类问题,有效利用标签之间的相关性是提高多标签分类模型精度的关键。针对该问题提出了高血压患者常见合并症的预测模型AR-MLKNN(multi-label k-nearest neighbor based on association rules),首先从不同语...合并症预测是典型的多标签分类问题,有效利用标签之间的相关性是提高多标签分类模型精度的关键。针对该问题提出了高血压患者常见合并症的预测模型AR-MLKNN(multi-label k-nearest neighbor based on association rules),首先从不同语义空间的临床概念中构建了患者特征表示,然后通过疾病标签关联信息量化合并症并发关系,并基于样本k邻域内标签的概率分布以后验概率的方式计算样本对每个疾病标签的隶属概率。利用合并症并发关系和疾病标签隶属概率映射形成合并症风险矩阵,基于合并症风险值,根据最小化分类损失的原则动态调整分类阈值以获取最优分类结果。实验结果表明该模型可以对高血压合并症进行较为准确的预测,F1-score达到82%,相较于常规的ML-KNN(multi-label k-nearest neighbor)模型提高了8%,在临床辅助决策领域具有一定的应用价值。展开更多
近年来,人们逐渐认识到临床大数据的潜在价值,疾病预测模型也开始成为临床研究的热点。不同种类疾病的预测模型在个体风险评估中发挥着越来越重要的作用。但由于一直没有针对疾病预测模型研究的报告规范,相关研究的报告结构和报告质量...近年来,人们逐渐认识到临床大数据的潜在价值,疾病预测模型也开始成为临床研究的热点。不同种类疾病的预测模型在个体风险评估中发挥着越来越重要的作用。但由于一直没有针对疾病预测模型研究的报告规范,相关研究的报告结构和报告质量大多参差不齐。2015年BMJ发表了《Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis(TRIPOD):the TRIPOD statement》,即TRIPOD声明,对疾病诊断和预后的预测模型研究的报告做了统一的规范。本文就TRIPOD声明的重点内容进行解读,以期促进理解和使用该报告规范。展开更多
The inflammatory bowel diseases(IBD),Crohn's disease(CD) and ulcerative colitis(UC),may be complicated by colorectal cancer(CRC).In a recent populationbased cohort study of 47 347 Danish patients with IBD by Tine ...The inflammatory bowel diseases(IBD),Crohn's disease(CD) and ulcerative colitis(UC),may be complicated by colorectal cancer(CRC).In a recent populationbased cohort study of 47 347 Danish patients with IBD by Tine Jess and colleagues 268 patients with UC and 70 patients with CD developed CRC during 30 years of observation.The overall risk of CRC among patients with UC and CD was comparable with that of the general population.However,patients diagnosed with UC during childhood or as adolescents,patients with long duration of disease and those with concomitant primary sclerosing cholangitis were at increased risk.In this commentary,we discuss the mechanisms underlying carcinogenesis in IBD and current investigations of genetic susceptibility in IBD patients.Further advances will depend on the cooperative work by epidemiologist and molecular geneticists in order to identify genetic polymorphisms involved in IBD-associated CRC.The ultimate goal is to incorporate genotypes and clinical parameters into a predictive model that will refine the prediction of risk for CRC in colonic IBD.The challenge will be to translate these new findings into clinical practice and to determine appropriate preventive strategies in order to avoid CRC in IBD patients.The achieved knowledge may also be relevant for other inflammation-associated cancers.展开更多
Background Resistance to anti-platelet therapy is detrimental to patients. Our aim was to establish a predictive model for aspirin resistance to identify high-risk patients and to propose appropriate intervention. Met...Background Resistance to anti-platelet therapy is detrimental to patients. Our aim was to establish a predictive model for aspirin resistance to identify high-risk patients and to propose appropriate intervention. Methods Elderly patients （n = 1130） with stable chronic coronary heart disease who were taking aspirin （75 mg） for 〉 2 months were included. Details of their basic characteristics, laboratory test results, and medications were collected. Logistic regression analysis was performed to establish a predictive model for aspirin resistance. Risk score was finally established according to coefficient B and type of variables in logistic regression. The Hosmer-Lemeshow （HL） test and receiver operating characteristic curves were performed to respectively test the calibration and discrimination of the model. Results Seven risk factors were included in our risk score. They were serum creatinine （〉 110 μmol/L, score of 1）; fasting blood glucose （〉 7.0 mmol/L, score of 1）; hyperlipidemia （score of 1）; number of coronary arteries （2 branches, score of 2; 〉 3 branches, score of 4）; body mass index （20-25 kg/m2, score of 2; 〉 25 kg/m2, score of 4）; percutaneous coronary intervention （score of 2）; and smoking （score of 3）. The HL test showed P ≥ 0.05 and area under the receiver operating characteristic curve ≥ 0.70. Conclusions We explored and quantified the risk factors for aspirin resistance. Our predictive model showed good calibration and discriminative power and therefore a good foundation for the further study of patients undergoing anti-platelet therapy.展开更多
文摘合并症预测是典型的多标签分类问题,有效利用标签之间的相关性是提高多标签分类模型精度的关键。针对该问题提出了高血压患者常见合并症的预测模型AR-MLKNN(multi-label k-nearest neighbor based on association rules),首先从不同语义空间的临床概念中构建了患者特征表示,然后通过疾病标签关联信息量化合并症并发关系,并基于样本k邻域内标签的概率分布以后验概率的方式计算样本对每个疾病标签的隶属概率。利用合并症并发关系和疾病标签隶属概率映射形成合并症风险矩阵,基于合并症风险值,根据最小化分类损失的原则动态调整分类阈值以获取最优分类结果。实验结果表明该模型可以对高血压合并症进行较为准确的预测,F1-score达到82%,相较于常规的ML-KNN(multi-label k-nearest neighbor)模型提高了8%,在临床辅助决策领域具有一定的应用价值。
文摘近年来,人们逐渐认识到临床大数据的潜在价值,疾病预测模型也开始成为临床研究的热点。不同种类疾病的预测模型在个体风险评估中发挥着越来越重要的作用。但由于一直没有针对疾病预测模型研究的报告规范,相关研究的报告结构和报告质量大多参差不齐。2015年BMJ发表了《Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis(TRIPOD):the TRIPOD statement》,即TRIPOD声明,对疾病诊断和预后的预测模型研究的报告做了统一的规范。本文就TRIPOD声明的重点内容进行解读,以期促进理解和使用该报告规范。
文摘The inflammatory bowel diseases(IBD),Crohn's disease(CD) and ulcerative colitis(UC),may be complicated by colorectal cancer(CRC).In a recent populationbased cohort study of 47 347 Danish patients with IBD by Tine Jess and colleagues 268 patients with UC and 70 patients with CD developed CRC during 30 years of observation.The overall risk of CRC among patients with UC and CD was comparable with that of the general population.However,patients diagnosed with UC during childhood or as adolescents,patients with long duration of disease and those with concomitant primary sclerosing cholangitis were at increased risk.In this commentary,we discuss the mechanisms underlying carcinogenesis in IBD and current investigations of genetic susceptibility in IBD patients.Further advances will depend on the cooperative work by epidemiologist and molecular geneticists in order to identify genetic polymorphisms involved in IBD-associated CRC.The ultimate goal is to incorporate genotypes and clinical parameters into a predictive model that will refine the prediction of risk for CRC in colonic IBD.The challenge will be to translate these new findings into clinical practice and to determine appropriate preventive strategies in order to avoid CRC in IBD patients.The achieved knowledge may also be relevant for other inflammation-associated cancers.
文摘Background Resistance to anti-platelet therapy is detrimental to patients. Our aim was to establish a predictive model for aspirin resistance to identify high-risk patients and to propose appropriate intervention. Methods Elderly patients （n = 1130） with stable chronic coronary heart disease who were taking aspirin （75 mg） for 〉 2 months were included. Details of their basic characteristics, laboratory test results, and medications were collected. Logistic regression analysis was performed to establish a predictive model for aspirin resistance. Risk score was finally established according to coefficient B and type of variables in logistic regression. The Hosmer-Lemeshow （HL） test and receiver operating characteristic curves were performed to respectively test the calibration and discrimination of the model. Results Seven risk factors were included in our risk score. They were serum creatinine （〉 110 μmol/L, score of 1）; fasting blood glucose （〉 7.0 mmol/L, score of 1）; hyperlipidemia （score of 1）; number of coronary arteries （2 branches, score of 2; 〉 3 branches, score of 4）; body mass index （20-25 kg/m2, score of 2; 〉 25 kg/m2, score of 4）; percutaneous coronary intervention （score of 2）; and smoking （score of 3）. The HL test showed P ≥ 0.05 and area under the receiver operating characteristic curve ≥ 0.70. Conclusions We explored and quantified the risk factors for aspirin resistance. Our predictive model showed good calibration and discriminative power and therefore a good foundation for the further study of patients undergoing anti-platelet therapy.
