Yu Xinzhu, Lophatananon Artitaya, Holmes Vivien, Muir Kenneth R, Guo Hui
Centre for Biostatistics, Division of Population Health, Health Services Research & Primary Care, School of Health Sciences, Faculty of Biology, Medicine and Health, The University of Manchester, Oxford Road, Manchester, M13 9PL UK.
Department of Epidemiology and Biostatistics, School of Public Health, Imperial College London, Sir Michael Uren Building, White City Campus, London, W12 0B UK.
NPJ Dement. 2025;1(1):4. doi: 10.1038/s44400-025-00006-2. Epub 2025 May 9.
Comprehensively studying modifiable risk factors to understand their contributions to dementia mechanisms is imperative. This study used natural language processing (NLP) models to pre-select candidate risk factors for dementia from 5505 baseline variables in the UK Biobank. We then applied causal discovery approaches to examine the relationships among the selected variables and their links to dementia in later life, presenting these connections in a causal network. We identified eight risk factors that directly or indirectly influence dementia, with mental disorders due to brain dysfunction (ICD-10 F06) acting as direct causes and mediators in pathways from other neurological disorders to dementia. Although evidence for the direct link between biological age and dementia was less pronounced, its potential value in dementia management remains non-negligible. This study advances our understanding of dementia mechanisms and highlights the potential of NLP and machine learning for the causal discovery of complex diseases from high-dimensional data.
全面研究可改变的风险因素以了解它们对痴呆症发病机制的影响至关重要。本研究使用自然语言处理（NLP）模型从英国生物银行的5505个基线变量中预先筛选痴呆症的候选风险因素。然后，我们应用因果发现方法来检验所选变量之间的关系及其与晚年痴呆症的联系，并将这些联系呈现为一个因果网络。我们确定了八个直接或间接影响痴呆症的风险因素，其中脑功能障碍所致精神障碍（ICD-10 F06）在从其他神经系统疾病到痴呆症的途径中作为直接原因和中介因素。尽管生物学年龄与痴呆症之间直接联系的证据不那么明显，但其在痴呆症管理中的潜在价值仍然不可忽视。本研究增进了我们对痴呆症发病机制的理解，并突出了NLP和机器学习从高维数据中进行复杂疾病因果发现的潜力。
