基于盆底超声参数构建的列线图模型预测盆底重建术后进展性压力性尿失禁的价值

目的 基于术前盆底超声参数构建列线图模型，探讨其预测盆底脏器脱垂（POP）患者盆底重建术后发生进展性压力性尿失禁（SUI）的价值。方法 选取2019年5月至2022年4月我院因国际泌尿科学会POP-Q分度≥Ⅱ度行盆底重建术女性患者495例建立预测模型作为模型组，其中术前POP合并SUI 295例；根据术前、术后SUI发生情况分为术后进展组76例（包括新发46例，加重30例）与SUI改善组219例；对模型组样本采取Bootstrap法重复抽样1000次作为内部验证组。应用单因素分析、多因素Logistic回归分析POP患者盆底重建术后发生进展性SUI的独立危险因素，并构建列线图模型。绘制受试者工作特征曲线、校准曲线、临床决策曲线、临床影响曲线对该模型区分度、校准度及临床适用性进行验证。结果 (1)单因素分析显示，SUI改善组与术后进展组孕次、产次、Ba点及指压诱发试验阳性、高血压史、糖尿病史、术前混合性尿失禁史、联合尿道中段无张力悬吊术（TVT-O）占比，以及盆底超声所测残余尿量（PVR）、尿道旋转角（Uret.rot）、尿道内口呈漏斗样改变、肛提肌损伤比较差异均有统计学意义（均P<...     

基于Bagging结合LASSO回归的规则集成方法对骨质疏松症患者肾阳虚证的辨识研究

目的：研究骨质疏松症（OP）患者的肾阳虚证辨识，形成中医临床辨证规则。方法：纳入982例OP患者的基本信息、病因病机、临床症状等特征信息，通过统计检验筛选数据中与肾阳虚证相关的变量。以决策树为基础模型，应用引导聚集算法（Bagging算法）建立OP肾阳虚证分类模型，生成大量规则并去除冗余；结合最小绝对值收敛和选择算子（LASSO）回归筛选关键规则并集成规则建立辨识模型，实现对OP患者肾阳虚证的辨识。结果：筛选出了18条关键辨识规则，其中11条规则的回归系数>0，对辨识为肾阳虚证具有正向影响，系数最高的规则为畏寒（“有”）&手足心热（“无”）；7条规则的回归系数<0，对辨识为肾阳虚证具有负向影响，系数最低的规则为舌红（“有”）&大便溏（“无”）&禀赋不足（“无”）。根据各关键规则的回归系数计算得到重要性>0.2的变量依次为畏寒、舌红、手足心热、肢冷、小便清、大便溏、禀赋不足、久病。辨识模型的偏依赖性分析结果显示，畏寒取值为“有”相较取值为“无”的OP患者，被辨识为肾阳虚证的概率高0.266 8，该变量对辨识为肾阳虚证具有最高影响。舌红取值为“有”...     

血清维生素A及碱性成纤维细胞生长因子水平预测早产儿视网膜病变的临床价值

目的：探讨血清维生素A(Vit A)及碱性成纤维细胞生长因子(bFGF)水平预测早产儿视网膜病变(ROP)的临床价值。方法：采用前瞻性队列研究。选取2020-01/2022-12在上海交通大学医学院附属上海儿童医学中心海南医院分娩且胎龄小于37 wk或出生体质量小于2 500 g的早产儿或低出生体质量儿411例作为研究对象，分别于出生后7 d和35 d进行外周血Vit A和bFGF水平检测。结果：最终完成临床研究的早产儿或低出生体质量儿共392例，其中1-2期ROP组 51例，3-5期ROP组 23例，未发生ROP组318例。生后7 d，1-2期ROP组血清Vit A(0.44±0.17 μmol/L)和bFGF(0.53±0.16 ng/L)水平分别低于未发生ROP组(0.50±0.12 μmol/L和0.63±0.15 ng/L)(均P&#x003C;0.05)； 3-5期ROP组血清Vit A(0.34±0.18 μmol/L)和bFGF(0.44±0.18 ng/L)水平分别低于未发生ROP组(均P&#x003C;0.05)； 3-5期ROP组血清Vit A和bFGF水平分别低于1-2期ROP组(均P&#x003C;0.05)。生后35d，3-5期ROP组血清Vit A(0.33±0.19 μmol/L)和bFGF(0.39±0.19 ng/L)水平分别低于1-2期ROP组(0.43±0.16 μmol/L和0.48±0.17 ng/L)(均P&#x003C;0.05)； 根据血清Vit A绘制的ROC曲线，AUC值为0.853，Youden指数最大值为0.68，最佳灵敏度为73%，最佳特异度为95%； 根据血清bFGF绘制的ROC曲线，AUC值为0.828，Youden指数最大值为0.58，最佳灵敏度为90%，最佳特异度为68%； 根据血清Vit A联合bFGF绘制的ROC曲线，AUC值为0.917，Youden指数最大值为0.70，最佳灵敏度为70%，最佳特异度为100%。结论：血清Vit A及bFGF水平均是预测ROP较敏感的有效指标，如果早产儿或低出生体质量儿血清Vit A或bFGF水平越低，可能提示ROP患病率越高及其病理分期越重。血清Vit A联合bFGF诊断ROP的临床价值较其单独诊断价值高，且减少误诊率。     

2010—2020年深圳市宝安区两街道流行性感冒流行特征及预测分析

目的:描述2010—2020年深圳市宝安区两街道流行性感冒的流行特征,并构建时间序列ARIMA模型,预测2021年流感发病趋势,为流感的防控提供科学依据。方法:采用stata15.0SE软件对数据进行整理与分析,发病率比较采用χ~2检验,并用自回归移动平均缓和模型(ARIMA),预测2021年的流感发病趋势。结果:2010—2020年深圳市宝安区两街道流感共5 179例病例,年均发病率为50.24/10万,2019年流感呈暴发流行(1 380例,发病率为125.45/10万);流感的年均发病率呈周期性,3月份发病数居多(811例,15.6%)。女性发病率(2 433例,55.38/10万)高于男性(2 756例,49.49/10万,χ~2=14.875,P0.05);人群分类主要以工人(1 048例,20.20%)、托幼儿童(958例,18.40%)为主;流感在各年龄均有发病,主要集中在0～5岁年龄段(1 586例,30.60%),重点防控场所中托幼机构的发病率最高(年均发病率6.22/10万,χ~2=1 210.808,P0.05)。ARIMA模型预测显示,2021年流感发病率与2020年相比整体呈下降趋势。结论:2021年深圳市宝安区两街道的流感发病率呈下降趋势,流感重点防控人群为工人及0～5岁儿童,建议普及流感疫苗接种,加强对工厂和托幼机构等重点场所的流感监测及监管。     

Improving variant prioritization in exome analysis by entropy-weighted ensemble of multiple tools

Variant prioritization is a crucial step in the analysis of exome and genome sequencing. Multiple phenotype-driven tools have been developed to automate the variant prioritization process, but the efficacy of these tools in clinical setting with fuzzy phenotypic information and whether ensemble of these tools could outperform single algorithm remains to be assessed. A large rare disease cohort with heterogeneous phenotypic information, including a primary cohort of 1614 patients and a replication cohort of 1904 patients referred to exome sequencing, were recruited to assess the efficacy of variant prioritization and their ensemble. Three freely available tools—Exomiser, Xrare, and DeepPVP—and their ensemble were evaluated. The performance of all three tools was influenced by the attributes of phenotypic input. When combining these three tools by weighted-sum entropy method (EWE3), the ensemble outperformed any single algorithm, achieving a rate of 78% diagnostic variants in top 3 (13% improvement over current best performer, compared to Exomiser: 63%, Xrare: 65%, and DeepPVP: 51%), 88% in top 10 and 96% in top 30. The results were replicated in another independent cohort. Our study supports using entropy-weighted ensemble of multiple tools to improve variant prioritization and accelerate molecular diagnosis in exome/genome sequencing.     

基于机器学习的脑卒中风险预测模型比较研究

目的：探讨多种机器学习预测模型对脑卒中发病风险的评估效果。方法：选取2013年1月1日—2017年12月31日参与“北京健康管理队列”的体检人群作为研究对象，基线人群共计56 017例。比较研究对象脑卒中发病与未发病人群之间基本人口学信息、代谢异常相关指标的差异，选用经典决策树模型、多层感知器模型、卷积神经网络模型开展模型构建，并与多因素logistic回归分析模型进行比较。结果：各模型分析结果均显示年龄、收缩压、腰围、身体质量指数为脑卒中发病的影响因素；多因素logistic回归分析模型、经典决策树模型、多层感知器模型、卷积神经网络模型的准确率分别为0.978、 0.985、 0.988、0.996。结论：代谢异常指标中的腹型肥胖、血压升高、低密度脂蛋白胆固醇降低、血糖升高均是脑卒中发病的潜在危险因素；经典决策树模型、多层感知器模型、卷积神经网络模型3种机器学习模型较多因素logistic回归分析模型预测性能更优，其中卷积神经网络模型的准确率最为良好，多层感知器模型的特异度最为良好。     

Chaotic genetic algorithm and Adaboost ensemble metamodeling approach for optimum resource planning in emergency departments

Long length of stay and overcrowding in emergency departments (EDs) are two common problems in the healthcare industry. To decrease the average length of stay (ALOS) and tackle overcrowding, numerous resources, including the number of doctors, nurses and receptionists need to be adjusted, while a number of constraints are to be considered at the same time. In this study, an efficient method based on agent-based simulation, machine learning and the genetic algorithm (GA) is presented to determine optimum resource allocation in emergency departments. GA can effectively explore the entire domain of all 19 variables and identify the optimum resource allocation through evolution and mimicking the survival of the fittest concept. A chaotic mutation operator is used in this study to boost GA performance. A model of the system needs to be run several thousand times through the GA evolution process to evaluate each solution, hence the process is computationally expensive. To overcome this drawback, a robust metamodel is initially constructed based on an agent-based system simulation. The simulation exhibits ED performance with various resource allocations and trains the metamodel. The metamodel is created with an ensemble of the adaptive neuro-fuzzy inference system (ANFIS), feedforward neural network (FFNN) and recurrent neural network (RNN) using the adaptive boosting (AdaBoost) ensemble algorithm. The proposed GA-based optimization approach is tested in a public ED, and it is shown to decrease the ALOS in this ED case study by 14%. Additionally, the proposed metamodel shows a 26.6% improvement compared to the average results of ANFIS, FFNN and RNN in terms of mean absolute percentage error (MAPE).