Cardiotoxicity is a common complication during the cancer treatment of breast cancer patients, increasing their physical and mental suffering and even endangering their lives. Therefore, early identification of high-risk patients and taking targeted preventive measures is crucial. Risk prediction models for cardiotoxicity related to breast cancer treatment are important methods for assessing high-risk patients and have been extensively studied by scholars both domestically and internationally. The paper categorizes existing models into specific and general models based on the treatment methods received by the study population. It provides a detailed explanation and comparative analysis of the construction processes, methods, predictive factors, validation of effectiveness, and applications of these models. The analysis shows that the construction methods of existing models mainly use traditional statistical methods such as Logistic regression and Cox proportional regression, with a few using machine learning algorithms. The predictive factors are mostly age, hypertension, diabetes, and BMI, with good predictive performance. However, there are significant differences in the predictive factors among the models, and there is an imbalance between development and application. Future research should focus on clinical application and localization verification of models, combine artificial intelligence technology, enrich new modeling methods, and conduct multi-center, large-sample prospective studies to improve model stability. This will provide effective screening tools for clinical healthcare professionals to identify cardiotoxicity. This paper can serve as a reference for the construction and application of risk prediction models for cardiotoxicity related to breast cancer treatment by clinical healthcare professionals in China.

心脏毒性是乳腺癌患者癌症治疗过程中常见并发症，增加了患者身心痛苦，甚至可能危及生命，故早期识别高危患者并采取针对性预防措施至关重要。乳腺癌治疗相关心脏毒性风险预测模型是评估高危患者的重要方法，且已被国内外学者广泛研究。本文根据研究对象所接受的治疗方法，将现有模型分为特异性模型和普适性模型，并对各模型的构建过程与方法、预测因子、效果验证与应用等内容进行了详细阐述和比较分析。分析表明现有模型的构建方法主要采用Logistic回归和Cox比例回归等传统统计方法，少数使用机器学习算法，预测因子多为年龄、高血压、糖尿病、BMI、蒽环类药物、曲妥珠单抗和紫杉醇，预测性能良好，但各模型预测因子差异较大，且呈现重开发而轻应用的不平衡局面。未来研究应注重模型临床应用和本土化验证，结合人工智能技术，丰富新型建模方法，开展多中心、大样本前瞻性研究，提高模型稳定性，为临床医护人员识别心脏毒性提供有效筛查工具。本文能够为我国临床医护人员构建及应用乳腺癌治疗相关心脏毒性风险预测模型提供借鉴。