trajectories
Domain A: Description of origins and processes of loss and recovery of control and modulating factors
In this domain, we will assess the trajectories of loss and regain of control and how triggers and modulating factors interact to predict when individuals with mild to moderate alcohol use disorder will lose or regain control of their consumption. We will identify general, but also age- and gender-specific, mechanisms that contribute to loss and regain of alcohol control by capturing a large cohort of individuals with varying degrees of severity who do not (yet) need detoxification. To do this, we will use ambulatory assessment methods (including ecological momentary assessment) to obtain longitudinal datasets over several months from people with an alcohol use disorder (with and without additional tobacco and cannabis use). This allows us to observe longitudinal trajectories of loss and regain of control over drug use in real life settings.
A01: Prospective longitudinal study on triggers and moderators in addiction disorders
The primary aim of the prospective longitudinal study is to investigate interindividual differences in predictive factors of trajectories of loss and regain of control over alcohol use across the lifespan. Using outpatient assessments, we will follow interindividual differences in behaviour, neurobiological correlates and polygenic risk scores in adolescents and young adults (N = 300; WP1) as well as early (N = 300; WP2) and late middle-aged (N = 300; WP3) patients with alcohol use disorder over one year. Using latent and mixed growth curve models, we will thus identify inter-individual differences and profiles regarding the prediction of maintained and lost control behaviours over alcohol use, which will help to critically inform new and person-centred prevention and intervention strategies. Project lead: Prof. Dr Dr Tobias Banaschewski, Mannheim Medical Faculty Prof. Dr Dr Andreas Heinz, Charité - Universitätsmedizin Berlin Prof Dr Dr Michael Rapp, University of Potsdam Project team members: Nadja Samia Bahr, Charité - Universitätsmedizin Berlin Friederike Deeken, University of Potsdam Marie Heigert, Charité - Universitätsmedizin Berlin Dr Patricia Pelz, Charité - Universitätsmedizin Berlin Julia Wenzel, Charité - Universitätsmedizin Berlin
A02: Longitudinal monitoring of cognitive control as a modifier of drinking behavior
The implicit assumption that intra-individual changes in cognitive control and decision-making behaviour are associated with changes in drinking behaviour has not yet been systematically investigated. This project aims to identify differential cognitive trajectories associated with the loss and regain of control over drinking behaviour. A smartphone application will be used to investigate cognitive control and decision-making behaviour longitudinally in outpatient assessment. Project leader: Prof. Dr Lorenz Deserno, University Hospital Würzburg Prof Dr Michael Smolka, Dresden University of Technology Project team members: Ying Lee, Dresden University of Technology Dr Hilmar Zech, Dresden University of Technology
A03: Stress-associated addiction pressure and relapse predictors in alcohol dependence
The project investigates the interaction of stress- and stimulus-induced physiological markers in a laboratory experiment with regard to: (1) addiction pressure and associated factors, (2) their suitability for relapse prediction in the living environment, and (3) the identification of brain regions associated with the relevant motivational, cognitive and affective processes. The longer-term goal is the selection of a set of mobile sensors and their integration into a mobile infrastructure for the prediction of stress-associated relapse situations in everyday life. Project management: Prof Dr Falk Kiefer, Central Institute of Mental Health Mannheim Prof Dr Clemens Kirschbaum, Dresden University of Technology Prof Dr Jan Stallkamp, Fraunhofer Institute for Manufacturing Engineering and Automation, Mannheim Project staff: Dr Patrick Bach, Central Institute of Mental Health Mannheim Dr Jens Langejürgen, Fraunhofer Institute for Manufacturing Engineering and Automation Philipp Radler, Fraunhofer Institute for Manufacturing Engineering and Automation Prof. Dr Sabine Vollstädt-Klein, Central Institute of Mental Health Mannheim Judith Zaiser, Central Institute of Mental Health Mannheim Sina Zimmermann, Central Institute of Mental Health Mannheim
A04: Intensive investigation of triggers, moderators and neurobiological mechanisms at transitions in addiction disorders
This combined outpatient assessment and imaging study identifies intra-individual differences in the triggers, moderators and mediating neurobiological mechanisms of dependence disorders at transitions with substantially increased or reduced alcohol consumption. Using high-frequency ambulatory assessment in everyday life, phase-dependent differences in the effects of triggers and moderators on alcohol craving, mood, impulsivity and drinking behaviour are investigated and the neurobiological correlates of these processes are identified using functional imaging. The prospective predictive value of mechanisms for maintained and lost control behaviour over alcohol consumption will be tested to critically inform new and person-centred prevention and intervention strategies. Project leader: Prof. Dr Ulrich Ebner-Priemer, Karlsruhe Institute of Technology Prof Dr Christine Heim, Charité - Universitätsmedizin Berlin Prof. Dr Dr Heike Tost, Central Institute of Mental Health Mannheim Project staff: Dr Gabriela Gan, Central Institute of Mental Health Mannheim Sarah Lohr, Central Institute of Mental Health Mannheim Ren Ma, Central Institute of Mental Health Mannheim Mirjam Melzer, Central Institute of Mental Health Mannheim
A05: High-resolution behavioral observations to capture dynamic transitions in animal models of addiction
In this project, we use new methods from statistical physics and dynamic systems to detect early warning signs and critical transitions to addictive behaviour. For this purpose, we use so-called intensive longitudinal data sets, which we collect with high-resolution automated behavioural observations in an animal model of alcohol addiction and in an animal model with voluntary long-term application of nicotine. Project leader: PD Dr Dr Hamid Noori, Max Planck Institute for Biological Cybernetics Tübingen, Central Institute of Mental Health Mannheim Prof Dr Rainer Spanagel, Central Institute of Mental Health Mannheim
A06: AI-based predictive neuro-behavioural modeling of individual trajectories in addiction
Our project aims to determine recurrent neural network models of behavioural dynamics from multimodal (mobile health) data on a person-specific level. To this end, we first develop methods that allow us to integrate data with different collection rates and distribution assumptions. This approach will be applied to data from projects A01- A04 to identify dynamic transitions between loss and regain of control, predict long-term behavioural trajectories and classify subgroups. Ultimately, we aim to determine the factors and mechanisms underlying these transitions. Project leader: Prof. Dr rer. nat. Daniel Durstewitz, Central Institute of Mental Health Mannheim Dr sc. hum. Georgia Koppe, Central Institute of Mental Health Mannheim
A07: Deep Learning to Identify Subtypes of Addiction Disorders Based on Structural MRI Data
In this subproject, we will develop explainable machine learning methods to systematically analyse structural MRI images with respect to sociodemographic variables, disease diagnosis, alcohol consumption and possible subtypes. In particular, we will build convolutional neural network (CNN) architectures and test them for (1) differential diagnosis of addicted patients and healthy controls (2) prediction of past and future alcohol consumption (3) determination of addiction subtypes and (4) general characterisation of mental health. In addition, we will visualise individual network decisions and integrate the CNN models with other ‘one-shot’ data from the TRR 265, such as socio-demographic and psychometric variables or newly found behavioural or fMRI biomarkers. Project management: Prof Dr Kerstin Ritter, Charité - Universitätsmedizin Berlin
A08: Effects of longitudinal sex hormone fluctuations on loss and recovery of alcohol control over time: prediction, mechanisms and treatment targets
The primary aim of the prospective longitudinal study is to investigate interindividual differences in predictive factors of trajectories of loss and regain of control over alcohol use across the lifespan. Using outpatient assessments, we will follow interindividual differences in behaviour, neurobiological correlates and polygenic risk scores in adolescents and young adults (N = 300; WP1) as well as early (N = 300; WP2) and late middle-aged (N = 300; WP3) patients with alcohol use disorder over one year. Using latent and mixed growth curve models, we will thus identify inter-individual differences and profiles regarding the prediction of maintained and lost control behaviours over alcohol use, which will help to critically inform new and person-centred prevention and intervention strategies. Project lead: Prof. Dr Dr Tobias Banaschewski, Mannheim Medical Faculty Prof. Dr Dr Andreas Heinz, Charité - Universitätsmedizin Berlin Prof Dr Dr Michael Rapp, University of Potsdam Project team members: Nadja Samia Bahr, Charité - Universitätsmedizin Berlin Friederike Deeken, University of Potsdam Marie Heigert, Charité - Universitätsmedizin Berlin Dr Patricia Pelz, Charité - Universitätsmedizin Berlin Julia Wenzel, Charité - Universitätsmedizin Berlin
A09: Predicting drinking behavior from functional brain connectomics by harvesting
information from multiple sources
Loosing and regaining control over alcohol consumption relies on executive functions such as impulse control, which are related to the functional architecture of the brain at rest and during specific tasks. In this project, we will therefore use the static as well as dynamic functional connectivity of the brain at rest and during the stop-signal task to predict the ability to control alcohol consumption. We will use innovative artificial intelligence (AI) techniques, such as normative models and meta-matching, emphasizing both interpretability and predictive accuracy. We will use relevance maps of the connectome to test specific functional hypotheses. As part of the above transfer learning techniques, we will use large data sets, particularly from the FOR 1617 cohort (N≈400), the IMAGEN cohort (N≈2000), and the UK Biobank (N≈36,000) to optimize predictive accuracy within the TRR cohort, which is important for clinical applications. Covariates will be used to examine AUD subtypes.