Presented During:
Wednesday, June 25, 2025: 5:45 PM - 7:00 PM
Brisbane Convention & Exhibition Centre
Room:
M2 (Mezzanine Level)
Poster No:
1597
Submission Type:
Abstract Submission
Authors:
Clara Vetter1, Florian Eichin1, David Popovic2, Clara Weyer2, Katharine Chrisholm3, Lana Kambeitz-Ilankovic4, Joseph Kambeitz4, Linda Antonucci5, Stephan Ruhrmann6, Anita Riecher-Rössler7, Rachel Upthegrove8, Raimo Salokangas9, Jarmo Hietala9, Christos Pantelis10, Rebekka Lencer11, Eva Meisenzahl12, Stephen Wood13, Paolo Brambilla14, Stefan Borgwardt15, Alessandro Bertolino5, Peter Falkai2, Daniel Rueckert1, Nikolaos Koutsouleris2
Institutions:
1Munich Center for Machine Learning (MCML), Munich, Germany, 2University of Munich, Munich, Germany, 3University of Sussex, Brighton , United Kingdom, 4University of Cologne, Cologne, Germany, 5University of Bari, Bari, Italy, 6Universiy of Cologne, Cologne, Germany, 7University of Basel, Basel, Switzerland, 8University of Oxford, Oxford, United Kingdom, 9University of Turku, Turku, Finland, 10Orygen, Melbourne, Victoria, 11University of Muenster, Muenster, Germany, 12Univesity of Duesseldorf, Duesseldorf, Germany, 13University of Melbourne, Parkville, Victoria, 14University of Milan, Milan, Italy, 15University of Luebeck, Luebeck, Germany
First Author:
Clara Vetter
Munich Center for Machine Learning (MCML)
Munich, Germany
Co-Author(s):
Florian Eichin
Munich Center for Machine Learning (MCML)
Munich, Germany
Introduction:
Psychiatric disorders often emerge during adolescence and early adulthood, a critical neurodevelopmental period characterized by heightened vulnerability to environmental influences. Early identification of individual risk profiles and timely interventions are therefore essential to improve long-term psychosocial and occupational functioning. However, the heterogeneity of risk and protective factors and lack of reliable biomarkers complicate this task. Neurodevelopmental pathways, driven by genetic predisposition and environmental stressors, contribute to clinical symptoms, cognitive deficits, and psychosocial dysfunction.
This study introduces a novel unsupervised machine learning method, multiblock sparse partial least squares (MB-SPLS), to integrate genetic, neuroanatomical, and clinical data. The multimodal approach provides a more comprehensive understanding of the multi-layered complexity of early-stage psychiatric disorders. Here, we aim to identify clinically relevant risk signatures to predict longitudinal functioning trajectories.
Methods:
Minimally medicated participants (n=1280) were acquired from the longitudinal PRONIA study, encompassing healthy control individuals, individuals with recent onset of psychosis and depression, as well as clinical high-risk states for psychosis. Multimodal data included structural neuroimaging segmented into 259 regions of interest using Multi-atlas region Segmentation utilizing Ensembles (MUSE), polygenic risk scores (PRS) for psychiatric and related traits, as well as clinical, neurocognitive, and functioning measures collected at baseline, 9-month, and 18-month follow-ups. MB-SPLS was employed to iteratively identify shared signatures, or latent components (LC), across the six data domains at baseline. LCs capture shared dimensions of variation by optimizing feature weights to maximize inter-block covariance. Statistical significance of LCs was assessed via permutation testing, and feature stability via bootstrap resampling. Longitudinal functioning trajectories were clustered using longitudinal k-means, and multiclass classification models were developed with linear support vector machines (SVM). All models were trained and tested in a (repeated) nested cross-validation framework.
Results:
We found a six-domain risk signature linking psychotic diagnoses (F2) and, less strongly, mood (F3) and anxiety (F4) disorders to declines in functioning, deficits in social cognition and verbal learning, childhood experiences of emotional abuse, increased PRS for schizophrenia alongside reduced PRS for educational attainment, and widespread gray matter alterations. Longitudinal clustering revealed three patient subgroups for both social and role functioning trajectories: a non-impaired group, an improving group, and a declining group which could significantly be predicted based on the dimensions of the risk signature. XAI analyses showed that no or little decline in baseline functioning was the most critical predictor of the non-impaired group, while increased genetic risk, increased emotional childhood trauma experiences, and brain structural changes predominantly in regions of the auditory, fronto-temporal, and basal ganglia networks distinguished between the improving and declining group.
Conclusions:
The risk signature underscores the biological and psychosocial factors driving heterogeneity in functioning trajectories. Poor functioning may reflect a biologically driven vulnerability, emphasizing the need for tailored, domain-specific interventions to enhance resilience, mitigate risk, and improve long-term outcomes in early-stage psychiatric disorders.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 2
Genetics:
Genetic Modeling and Analysis Methods
Lifespan Development:
Early life, Adolescence, Aging
Modeling and Analysis Methods:
Classification and Predictive Modeling
Multivariate Approaches 1
Keywords:
Cognition
Computational Neuroscience
Data analysis
Machine Learning
Multivariate
Psychiatric Disorders
Schizophrenia
Statistical Methods
STRUCTURAL MRI
Other - Genetics, Childhood Trauma
1|2Indicates the priority used for review
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
I accept
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information.
Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
I do not want to participate in the reproducibility challenge.
Please indicate below if your study was a "resting state" or "task-activation” study.
Other
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Patients
Was this research conducted in the United States?
No
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel?
NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Yes
Were any animal research approved by the relevant IACUC or other animal research panel?
NOTE: Any animal studies without IACUC approval will be automatically rejected.
Not applicable
Please indicate which methods were used in your research:
Structural MRI
Behavior
Neuropsychological testing
Computational modeling
Other, Please specify
-
Genetics
For human MRI, what field strength scanner do you use?
1.5T
Which processing packages did you use for your study?
SPM
Other, Please list
-
CAT12
Provide references using APA citation style.
tba
No