This dissertation aimed to evaluate passive behavioural data as a new data source in the field of public mental health. To date, research in the field of public mental health has primarily used written or oral self-report data as a method, which requires great effort, is time-consuming, and is susceptible to report bias. The conceptualisation of prevention measures and needs in the space of public mental health requires data that are as up-to-date and precise as possible. The use of smartphone-assessed passive behavioural data has already been discussed as additional objective measurement method to identify symptoms of mental disorders. Standardised parameters of passive behavioural data do not yet exist within the public mental health context. However, before passive behavioural data can be integrated as a new data source in the field of public mental health research, it must be evaluated on the basis of quality criteria. The evaluation of passive behavioural data is accompanied by essential ethical (and data protection) questions. Due to the currently high prevalence of depressive disorders worldwide and their tremendous individual and social consequences, the evaluation of passive behavioural data in the present dissertation was carried out by using depressive symptoms as the example. The three research projects in this dissertation were based on data collection aiming to predict depressive symptoms with the help of passive behavioural data. The results from Research Project 1 suggested that passive behavioural smartphone data offered added content value compared to survey data of established self-reported predictors of depressive symptoms. Of the included passive behavioural smartphone data, messenger use and video calls, as correlates of social interaction, contributed significantly to the prediction of depressive symptoms. The results of Research Project 1 suggested that passive behavioural data could be used as measures of depressive symptoms; a precise attribution of passive behavioural data points to individual depressive symptoms is still pending. This would enable continuous data collection in the public mental health sector and thus early identification of trends in the prevalence of depressive symptoms. This, in turn, provides the basis for the equally timely planning and implementation of preventive measures. In addition, Research Project 2 investigated the contribution of regional data derived from official statistics based on GPS location to the prediction of individual depressive symptoms. The results showed that regional information could be helpful in identifying risk groups in the public mental health sector. The results were not transferable to the postpandemic period due to the survey being conducted during the COVID-19 pandemic but, at the same time, provide valuable information on groups particularly at risk for depressive symptoms during a pandemic. Research project 3 investigated the extent to which characteristics of passive behavioural data are valid as correlates of depressive symptoms across individuals. To this end, we tested the extent to which personality facets and perceived social support were associated with the relationship between media use and physical activity levels (measured as behavioural data) and depressive symptoms. Some behavioural data points were found to be related to depressive symptoms independently of individual characteristics, while others were related to depressive symptoms in differential ways. Thus, the results of Research Project 3 highlighted the relevance of interindividual differences in the assessment of passive behavioural data. Overall, the results indicated that defining consistent behavioural data measures for the public mental health field is challenging, as some behavioural data measures show different strengths of associations with depressive symptoms between individuals. As a practical implication, both an individual-centred approach and the inclusion of person-specific control variables are recommended.
