临床预测模型是通过将多因素纳入以估算发病或预后结局发生概率的工具,包括诊断模型和预后模型2种类型.可视化是将数据、信息和知识转化为可视形式的过程,用于更深入地展示数据分析的结果.临床预测模型结果的常用可视化方法包括列线图、网页计算器、评分系统、图形评分表以及机器学习中的 Shapley Additive Explanations(SHAP)和 Local Interpretable Model-Agnostic Ex-planations(LIME)方法,这些方法能够直观地展示预测结果,提高模型的可解释性.近年来,在中医药领域内亦兴起关于预测模型的相关研究.本研究旨在探讨临床预测模型结果的可视化方法在中医领域的应用现状,对临床预测模型结果的多种可视化方法进行汇总介绍,并对多种可视化方法进行对比,为选择合适的中医临床预测模型结果展示方法提供参考.
Abstract
Clinical prediction models are analytical tools designed to estimate the likelihood of clinical outcomes by integrating a multitude of factors.These encompass both diagnostic and prognostic elements.Visualization refers to the conversion of data and findings into a graphical format,enriching the depth of data analysis representation.Standard visualization techniques for the outcomes of clinical prediction models include bar and line charts,web-based calculators,scoring systems,graphical rating scales,and machine learning interpretative tools such as Shapley Additive Explanations(SHAP)and Local Interpretable Model-Agnostic Explanations(LIME).These techniques provide intuitive and insightful representations of predictive results,enhancing the comprehensibility of the models.Recently,predictive modeling has gained traction in the context of traditional Chinese medicine(TCM).This study seeks to explore the visualization practices of clinical prediction models in TCM,offering an overview and comparative analysis of various visualization methods.The goal is to aid in selecting an optimal approach for presenting clinical prediction outcomes in TCM.
维普
万方数据