OBJECTIVE: The objective of this paper is to investigate the goals and variables employed in the machine learning and microsimulation studies for the prognosis of dementia. METHOD: According to preset protocols, the Pubmed, Socups and Web of Science databases were searched to find studies that matched the defined inclusion/exclusion criteria, and then its references were checked for new studies. A quality checklist assessed the selected studies, and removed the low quality ones. The remaining ones (included set) had their data extracted and summarized. RESULTS: The summary of the data of the 37 included studies showed that the most common goal of the selected studies was the prediction of the conversion from mild cognitive impairment to Alzheimer's Disease, for studies that used machine learning, and cost estimation for the microsimulation ones. About the variables, neuroimaging was the most frequent used. CONCLUSIONS: The systematic literature review showed clear trends in prognosis of dementia research in what concerns machine learning techniques and microsimulation.
展开▼
机译：目的：本文的目的是研究在机器学习和微观模拟研究中用于痴呆预后的目标和变量。方法：根据预设方案，搜索Pubmed，Socups和Web of Science数据库，以查找符合定义的纳入/排除标准的研究，然后检查其参考文献以进行新的研究。质量检查表评估了选定的研究，并删除了低质量的研究。其余的（包括集合）将其数据提取并汇总。结果：纳入的37项研究的数据摘要表明，所选择研究的最普遍目标是针对使用机器学习的研究和从微观模拟的成本估算中预测轻度认知障碍向阿尔茨海默氏病的转化。关于变量，神经成像是最常用的。结论：系统的文献综述显示，痴呆症研究的预后趋势在机器学习技术和微观模拟方面具有明显趋势。
展开▼
