Cognitive impairment and dementia have dramatic individual and social consequences, and create high economic costs for societies. In order to delay cognitive aging of future generations as long as possible, we need evidence about which contextual factors are most supportive for individuals to reach highest cognitive levels relative to their potential. At the same time, for current older generations, we need scalable methods to exactly identify individuals at risk of cognitive impairment. The project intends to apply recent methodological and statistical advancements to reach two objectives. Firstly, contextual influences on cognitive aging will be comparatively assessed, with a focus on inequalities related to educational opportunities and gender inequalities. This will be done using longitudinal, population-representative, harmonized cross-national aging surveys, merged with contextual information. Secondly, the project will quantify the ability of singular and clustered individual characteristics, such as indicators of cognitive reserve and behaviour change, to predict cognitive aging and diagnosis of dementia. Project methodology will rely partly on parametric ‘traditional’ multilevel- or fixed-effects modelling, partly on non-parametric statistical learning approaches, to address objectives both hypothesis- and data-driven. Applying statistical learning techniques in the field of cognitive reserve will open new research avenues for efficient handling of large amounts of data, among which most prominently the accurate prediction of health and disease outcomes. Quantifying the role of contextual inequalities related to education and gender will guide policymaking in and beyond the project. Assessing risk profiles of individuals in relation to cognitive aging will support efficient and scalable risk screening of individuals. Identifying the value of behaviour change to delay cognitive impairment will guide treatment plans for individuals affected by dementia.