About the Project

Our Mission

The BulliedMind project aims to investigate multimodal neurobiological markers as potential mechanisms mediating the link between childhood bullying and mental health outcomes from childhood through adulthood. 

By using data from two major European population-based studies, the English Avon Longitudinal Study of Parents and Children (ALSPAC) and the German Bavarian Longitudinal Study (BLS), we strive to provide comprehensive insights into the development and long term impact of bullying on brain structure and conectivity, and its impact on mental health.

Key Objectives

• Examine the Impact of Bullying: We explore whether childhood and adolescent peer victimization leads to poorer mental health throughout adulthood. Our research focuses on a wide range of mental health measures, including internalizing and externalizing problems, emotional disorders, and behavioral issues.

• Investigate Neurobiological Pathways: Our project delves into the structural and functional changes within the brain, particularly focusing on intrinsic functional connectivity and structural networks like the default mode network (DMN) and salience network (SN). We aim to understand how these morphological and network changes mediate the relationship between early bullying and mental health outcomes.

• Identify Resilience via Machine Learning: Utilizing multimodal longitudinal data, we create data-driven clustering models to map complex interactions, uncovering individual differences and specific protective mechanisms that mitigate adverse outcomes.

Our Approach

The BulliedMind project employs a robust methodological framework, including:

• Cross-Dataset Analysis: We combine harmonized data from multiple cohorts to increase the power and generalizability of our findings, using the open R package VeryWise for pooled surface-based morphological analysis.

• Advanced Neuroimaging Techniques: Using structural MRI, resting-state fMRI, and diffusion tensor imaging (DTI) data, we analyze morphological brain changes as well as structural and intrinsic functional connectivity patterns associated with childhood bullying.

• Explainable Artificial Intelligence (XAI): We adopt non-linear machine learning models, such as deep recurrent neural networks and variational autoencoders, coupled with XAI frameworks like SHAP and LIME. This allows us to move beyond broad categorizations to capture personalized, individual-level insights into mental health trajectories and resilience