O'Dea Bridianne, Batterham Philip J, Braund Taylor A, Chakouch Cassandra, Larsen Mark E, Berk Michael, Torok Michelle, Christensen Helen, Glozier Nick
Flinders University Institute for Mental Health and Wellbeing, Flinders University, Adelaide, South Australia, Australia.
Black Dog Institute, University of New South Wales, Sydney, New South Wales, Australia.
Npj Ment Health Res. 2025 Jul 19;4(1):30. doi: 10.1038/s44184-025-00140-y.
Linguistic features within individuals' text data may indicate their mental health. This trial examined the linguistic markers of depressive and anxiety symptoms in adults. Using a randomised cross over trial design, 218 adults provided eight different types of text data of varying frequencies and emotional valance. Linguistic features were extracted using LIWC-22 and correlated with self-reported symptoms. Machine learning was used to determine associations. No linguistic features were consistently associated with depressive or anxiety symptoms within or across all tasks. Features associated with depressive symptoms were different for each task and there was only some degree of reliability of these features within tasks. In all machine learning models, predicted values were weakly associated with actual values. Some text tasks had lower levels of engagement and negative impacts on mood. Overall, the linguistic markers of depression and anxiety shifted in response to contextual factors and the nature of the text analysed. This trial was prospectively registered with the Australian New Zealand Clinical Trials Registry (date registered: 15 September 2021, ACTRN12621001248853).
个体文本数据中的语言特征可能表明其心理健康状况。本试验研究了成年人抑郁和焦虑症状的语言标记。采用随机交叉试验设计，218名成年人提供了八种不同类型、频率和情感效价各异的文本数据。使用LIWC-22提取语言特征，并将其与自我报告的症状进行关联。运用机器学习来确定关联关系。在所有任务中或跨所有任务，均没有语言特征与抑郁或焦虑症状始终存在关联。与抑郁症状相关的特征在每个任务中都有所不同，并且这些特征在任务中的可靠性仅处于一定程度。在所有机器学习模型中，预测值与实际值的关联较弱。一些文本任务的参与度较低，且对情绪有负面影响。总体而言，抑郁和焦虑的语言标记会根据上下文因素和所分析文本的性质而发生变化。本试验已在澳大利亚新西兰临床试验注册中心进行前瞻性注册（注册日期：2021年9月15日，ACTRN12621001248853）。
