Abstract:

ObjectiveTo systematically search the research literature related to the application of machine learning models in hospice care，with a view to providing references for clinical practice. Methods A systematic search of Wanfang database，CNKI，VIP database，China Biomedical Literature Database，PubMed，Embase，Scopus，Cochrane Library，Web of Science，and CINAHL was conducted in accordance with the methodology of the scoping review as a guideline，with the timeframe of searching from the establishment of the database to August 30，2024，and the included literature was screened，summarized，extracted，and analyzed. Results Totally 17 studies were included. Analysis revealed that supervised machine learning algorithms（including random forest，decision tree，and neural networks） predominated in palliative care applications. Data sources and collection methods varied widely，with models applied across diverse scenarios. Model functions include assessing hospice needs，predicting a patient's risk of death，assisting with symptom management，analyzing hospice communication content，and more. Conclusion Machine learning models in palliative care demonstrate considerable utility and broad applicability. Future research should enhance data quality，optimize model development workflows，and improve model performance.

Key words: Hospice Care, Machine Learning, Scoping Review, Nursing Care

摘要： 目的 系统回顾关于机器学习模型在安宁疗护中应用的文献，为相关模型的构建和临床应用提供参考。方法 以范围综述方法论为指导，系统检索Web of Science、PubMed、Scopus、Cochrane Library、Embase、CINAHL、中国知网、维普数据库、万方数据库和中国生物医学文献数据库，检索时限为建库至2024年8月30日，对检索到的文献进行筛选、汇总、提取、分析。结果 共纳入17篇文献。分析结果显示，安宁疗护机器学习模型中使用的机器学习算法以监督式机器学习算法为主，包括随机森林、决策树和神经网络等；模型的数据来源及资料收集方式多样，且应用场景广泛；模型功能包括评估安宁疗护需求、预测患者的死亡风险、协助症状管理、分析安宁疗护沟通内容等。结论 安宁疗护中机器学习模型的适用性较高、应用范围较广。未来研究应提高数据质量，优化模型的开发流程，提升模型性能。

关键词: 安宁疗护, 机器学习, 范围综述, 护理