预后研究(预测模型)质量评价(Prediction model Risk Of Bias ASsessment Tool, PROBAST)(二)

发布于 2024年8月30日星期五 23:43:31 浏览：429

原创不易，转载请注明来源，感谢！

在预后研究(预测模型)质量评价(Prediction model Risk Of Bias ASsessment Tool, PROBAST)(一)一文中介绍了临床预测模型的概念、PROBAST概述及使用方法。本文主要介绍PROBAST偏倚风险和适用性评价条目清单。研究对象领域评价条目清单解读、预测因素领域评价条目清单解读、结局领域评价条目清单解读、分析领域信号问题4.1-4.5评价条目清单解读、分析领域信号问题4.6-4.9评价条目清单解读和重要知识点详见其他推文。

关键词：诊断预测模型质量评价; 预后预测模型质量评价; 临床预测模型质量评价; 预测模型偏倚风险评估; PROBAST

一、研究对象

(一) 偏倚风险评估

Describe the sources of data and criteria for participant selection: 描述研究对象的纳排标准和数据来源：
Signaling questions 信号问题		Dev 训练集	Val 验证集
1.1 Were appropriate data sources used, e.g. cohort, RCT or nested case-control study data? 1.1数据来源是否合适，如队列研究，随机对照试验或巢式病例对照研究？
1.2 Were all inclusions and exclusions of participants appropriate? 1.2纳入和排除标准是否合理？
Risk of bias introduced by selection of participants 选择研究对象引入的偏倚风险	RISK: (low/high/unclear) 偏倚风险：(低/高/不清楚)
Rationale of bias rating: 偏倚风险分级的推理说明：

(二) 适用性评估

Describe included participants, setting and dates: 描述研究对象的纳排标准、招募背景及日期：
Concern that the included participants and setting do not match the review question 纳入的研究对象和背景与所关注的综述问题不匹配	CONCERN: (low/high/unclear) 风险：(低/高/不清楚)
Rationale of applicability rating: 适用性风险分级的推理说明：

二、预测因素

(一) 偏倚风险评估

List and describe predictors included in the final model, e.g. definition and timing of assessment: 列出并描述最终模型中包含的预测因子，包括其定义和测量时间：
Signaling questions 信号问题		Dev 训练集	Val 验证集
2.1 Were predictors defined and assessed in a similar way for all participants? 2.1所有研究对象的预测因子定义及测量是否一致或相似？
2.2 Were predictor assessments made without knowledge of outcome data? 2.2预测因子的测量是否与结局无关？
2.3 Are all predictors available at the time the model is intended to be used? 2.3在模型使用的时间点，是否能够得到所有预测因子的信息？
Risk of bias introduced by predictors or their assessment 预测因素引入的偏倚风险	RISK: (low/high/unclear) 偏倚风险：(低/高/不清楚)
Rationale of bias rating: 偏倚风险分级的推理说明：

(二) 适用性评估


Concern that the definition, assessment or timing of predictors in the model do not match the review question 预测因素的定义、测量方法及测量时间与所关注的综述问题不匹配	CONCERN: (low/high/unclear) 风险：(低/高/不清楚)
Rationale of applicability rating: 适用性风险分级的推理说明：

三、结局

(一) 偏倚风险评估

Describe the outcome, how it was defined and determined, and the time interval between predictor assessment and outcome determination: 对结局指标进行描述，包括定义和为什么确定这些指标，以及预测评估和确定结局之间的时间间隔：
Signaling questions 信号问题		Dev 训练集	Val 验证集
3.1 Was the outcome determined appropriately? 3.1结局的测量方法是否合适？
3.2 Was a pre-specified or standard outcome definition used? 3.2结局是否是采用预先设定的标准的定义？
3.3 Were predictors excluded from the outcome definition? 3.3结局的定义中是否排除了预测因子信息？
3.4 Was the outcome defined and determined in a similar way for all participants? 3.4所有研究对象的结局定义和测量是否一致或相似？
3.5 Was the outcome determined without knowledge of predictor information? 3.5结局的测定是否与预测因子无关？
3.6 Was the time interval between predictor assessment and outcome determination appropriate? 3.6预测因子的测量和确定结局之间的时间间隔是否合适？
Risk of bias introduced by the outcome or its determination 结局指标引入的偏倚风险	RISK: (low/high/unclear) 偏倚风险：(低/高/不清楚)
Rationale of bias rating: 偏倚风险分级的推理说明：

(二) 适用性评估

At what time point was the outcome determined: 在什么时间点确定结局：
If a composite outcome was used, describe the relative frequency/distribution of each contributing outcome: 如果使用了复合结局指标，请描述每个构成部分的相对频率/分布：
Concern that the outcome, its definition, timing or determination do not match the review question 结局指标的定义、测量方法及测量时间与所关注的综述问题不匹配	CONCERN: (low/high/unclear) 风险：(低/高/不清楚)
Rationale of applicability rating: 适用性风险分级的推理说明：

四、分析

(一) 偏倚风险评估

Describe numbers of participants, number of candidate predictors, outcome events and events per candidate predictor: 描述研究对象的数量、候选预测因子数量、结局事件数量和每个候选预测因子的结局事件发生情况：
Describe how the model was developed (for example in regards to modelling technique (e.g. survival or logistic modelling), predictor selection, and risk group definition): 描述模型是如何开发的(如关于建模方法[如生存模型或逻辑回归模型]、预测因素选择过程和风险组划分：
Describe whether and how the model was validated, either internally (e.g. bootstrapping, cross validation, random split sample) or externally (e.g. temporal validation, geographical validation, different setting, different type of participants): 描述是否以及如何对模型开展内部(如bootstrap法、交叉验证、随机分割样本验证)或外部(如时间外部验证、地区外部验证、不同环境、不同类型的研究对象)验证：
Describe the performance measures of the model, e.g. (re)calibration, discrimination, (re)classification, net benefit, and whether they were adjusted for optimism: 描述模型的性能指标，如校准度、区分度、重分类、净效益，以及是否对它们进行了乐观调整：
Describe any participants who were excluded from the analysis: 描述分析中排除的研究对象：
Describe missing data on predictors and outcomes as well as methods used for missing data: 描述预测因素和结局数据的缺失情况及处理方法：
Signaling questions 信号问题		Dev 训练集	Val 验证集
4.1 Were there a reasonable number of participants with the outcome? 4.1发生结局的研究对象数量是否合理？
4.2 Were continuous and categorical predictors handled appropriately? 4.2连续性变量和分类变量的处理是否合适？
4.3 Were all enrolled participants included in the analysis? 4.3分析中是否纳入了所有的研究对象？
4.4 Were participants with missing data handled appropriately? 4.4对缺失数据的处理是否合适？
4.5 Was selection of predictors based on univariable analysis avoided? 4.5是否避免了基于单因素分析选择预测因子？（仅适用于模型开发）
4.6 Were complexities in the data (e.g. censoring, competing risks, sampling of controls) accounted for appropriately? 4.6数据中的复杂问题(例如数据删失，竞争风险和对照组的抽样)是否被合理处理？
4.7 Were relevant model performance measures evaluated appropriately? 4.7模型性能的评估是否合适？
4.8 Were model overfitting and optimism in model performance accounted for? 4.8模型性能的过拟合、欠拟合是否被合理校正？ (仅适用于模型开发)
4.9 Do predictors and their assigned weights in the final model correspond to the results from multivariable analysis? 4.9最终模型中的预测因子及其权重是否与呈现出来的多因素分析结果相一致？ (仅适用于模型开发)
Risk of bias introduced by the analysis 分析引入的偏倚风险	RISK: (low/high/unclear) 偏倚风险：(低/高/不清楚)
Rationale of bias rating: 偏倚风险分级的推理说明：

本文内容是参考相关文献后对预后研究(预测模型)质量评价(Prediction model Risk Of Bias ASsessment Tool, PROBAST)工具的概述，仅代表本网站观点。关于PROBAST工具的更多内容可参考Robert F Wolff等发表的文章PROBAST: A Tool to Assess the Risk of Bias and Applicability of Prediction Model Studies，或Karel G M Moons等发表的文章PROBAST: A Tool to Assess Risk of Bias and Applicability of Prediction Model Studies: Explanation and Elaboration。

我要纠错

End

预后研究(预测模型)质量评价(Prediction model Risk Of Bias ASsessment Tool, PROBAST)(一)

预后研究(预测模型)质量评价(Prediction model Risk Of Bias ASsessment Tool, PROBAST)(三)