心血管风险预测的机器学习：临床医生的实用入门

摘要(中文译文)

机器学习（ML）为临床风险预测提供了强大的工具，通过改善模式识别和预测能力增强了传统统计方法——其在心血管护理和研究中的作用和应用正在不断扩大。本指南为读者提供用于风险预测的机器学习方法论基础知识，包括模型开发、验证和实施方面的考量。我们讨论了监督学习和无监督学习方法、特征选择、性能指标，以及深度学习中断时间序列分析等先进技术。在探讨可解释性、偏倚和临床整合等挑战的同时，也为读者在实践中评估或实施基于机器学习的风险预测工具提供了实用建议。本入门指南使读者具备必要的知识，以批判性地评估基于机器学习的风险预测模型，并与数据科学家有效合作。

Abstract

Machine learning (ML) offers powerful tools for clinical risk prediction, enhancing traditional statistical approaches through improved pattern recognition and predictive capability-its role and use in cardiovascular care and research is expanding. This guide provides readers with foundational knowledge of ML methodology for risk prediction, including model development, validation, and implementation considerations. We discuss supervised and unsupervised learning approaches, feature selection, performance metrics, and advanced techniques such as deep learning and interrupted time-series analysis. Challenges regarding interpretability, bias, and clinical integration are addressed alongside practical recommendations for readers evaluating or implementing ML-based risk prediction tools in practice. This primer equips readers with essential knowledge to critically appraise ML-based risk prediction models and collaborate effectively with data scientists.

如何引用

Heart, lung & circulation 的其他文献

• 心力衰竭和衰老临床前模型中心脏脆弱性的评估——发现与临床之间的平行关系