Methodological transitions in mental health research driven by machine learning

Machine learning is increasingly reshaping psychological and mental health research by complementing traditional theory-driven approaches with data-driven predictive modelling. This encompasses three key transitions: from hypothesis-testing to pattern discovery; from controlled experimental settings to naturalistic and multidimensional data ecosystems; and from explanatory theoretical models to prediction-informed theory-building and clinical translation. Nevertheless, this methodological reorientation remains insufficiently synthesised. Our aim is to examine how machine learning is reshaping psychological research methodologies, focusing on its ability to process and analyse multidimensional data, its capacity to model complex associations and temporal patterns and its integration into clinical practice. We critically examine both the opportunities and boundaries of integrating predictive models with theoretical insights. We further clarify that predictive accuracy, interpretability and causal explanation are related but distinct aims, and that machine learning should complement rather than replace psychological theory, causal reasoning and clinical judgement.