The second year of the Pattern-Cog project has just concluded. As we look back, it’s clear that our efforts have yielded significant advancements and milestones. Each Work Package (WP) has contributed significantly to our collective progress. Here are some of the key advances and major achievements from each WP. We would like to thank all partners for their impressive work and we look forward to the busy months ahead!
WP1: Extension of BrainAGE framework and development of MRI-derived local confidence maps
- The BrainAGE approach was refined and extended to include regional BrainAGE estimates, which allow the use for transdiagnostic purposes and were tested in different diseases. Details can be found in the recent paper “BrainAGE: Revisited and reframed machine learning workflow” by Kalc et al in the journal Human Brain Mapping (https://doi.org/10.1002/hbm.26632).
- Conceptual improvements related to prediction and clinical application of the BrainAGE approach were discussed and presented to the scientific community in an invited Perspective paper (Gaser et al., 2024, Nature Computational Science, accepted).
- The structural data from the Pattern-Cog database have been pre-processed and the processed data are currently being prepared for integration into other work packages.
WP2: Personalised ageing pattern framework based on multi-view learning
- Study prediction of future amyloid burden/positivity using ADNI data.
- Development of multi-view Machine Learning (ML) approaches for relevant prediction tasks in very early signs of dementia (e.g. MCI) and intervention.
- Findings showing that a model predicting amyloid conversion predicts also healthy-to-MCI conversion well (see Saadmaan et al paper published in Alzheimer’s & Dementia https://doi.org/10.1002/alz.13843).
- Study the role of noisy labels in the below tasks
- Proposal of a novel genetic algorithm-based approach for selecting optimal feature subsets in binary classification with noisy labels (see conference paper: https://doi.org/10.1007/978-3-031-56950-0_33)
- Integration of longitudinal data, plasma-markers, brain-AGE, and normative models into the prediction models.
WP3: High-performance algorithm for screening healthy individuals for risk of impending MCI and subsequent AD
- Looking at three algorithmic approaches to predict Mild Cognitive Impairment (MCI).
- Training of the models using different sets of features to ensure the interpretability of the models, and to clearly know where the effort in a future clinical setting should be done in order to ascertain if a patient will remain cognitively healthy in the future.
- Analyses showed that the most reliable features to determine if cognitively healthy individuals will lose their cognitive abilities in the long term are the thickness of the entorhinal cortex, the volume of the hippocampus, the memory performance, and some genetic risk factors.
WP4: Personalised ageing pattern framework in cognitive decline/dementia prevention RCTs
- 3059 participants were invited to participate in the MET-FINGER trial (i.e. 545 screened for lifestyle, 356 eligible and 184 randomized). Ethical approval for magnetic resonance imaging (MRI) was successful and all participants at Karolinska Institutet can undergo neuroimaging starting Q3 2024.
- Update of the Neuroimaging data for the FINGER trial with the new version of the FreeSurfer software, and calculation of BrainAge index.
- Testing of Alzheimer’s disease polygenic risk score and APOE4 genotype against neuroimaging in the FINGER multidomain intervention trial. APOE4 status was significantly associated with lower hippocampal volumes, and higher amyloid accumulation at baseline, and additionally with 2-year decline in brain volumes and thickness. Brain amyloid accumulation increased less in intervention group compared to control group in persons with higher genetic risk for Alzheimer´s disease (see Saadmaan et al. paper published in Alzheimer’s & Dementia https://doi.org/10.1002/alz.13843)
WP5: Ethical, legal and social aspects of personalised medicine in dementia prediction/prevention
Data aspects:
- Preparation of EBRAINS Health Data Cloud for the new Europrivacy Seal to demonstrate GDPR compliance
- Preparation of several human data sets for sharing via the Health Data Cloud including structuring data and data curation process.
- Preparation of new data sets that will be soon discoverable publicly via their non-sensitive meta data.
Patient and Public Involvement (PPI):
- Regular contact with the Pattern-Cog advisory board, set up by Alzheimer Europe in March 2023, which is composed of members of the European Dementia Carers Working Group (EDCWG).
- Organisation of three consultations with the Pattern-Cog Advisory Board.
- First consultation (face-to-face meeting, Helsinki, October 2023), following a previous meeting in Brussels in March 2023. This session focused on developing recommendations for clinicians on how to communicate algorithm-based risk predictions. The insights from this meeting were instrumental in shaping Alzheimer Europe’s position paper on risk disclosure, published in November 2023 (Alzheimer Europe calls for urgent action to disclose dementia risk in an open, honest, empathetic and compassionate way).
- Second consultation (online meeting, November 2023) focused on concerns and worries about the use of Artificial Intelligence in dementia research projects, such as Pattern-Cog.
- Third consultation (online meeting, April 2024) focused on addressing the needs of cognitively healthy people (with or without cognitive complaints) at risk of developing Alzheimer’s disease and how to deal with these needs, which is very relevant in the Pattern-Cog project and others looking at detecting signs indicative of future cognitive impairment.
WP6: Communication and dissemination
- Maintenance of the project’s communication tools including social media channels (X, LinkedIn) and website. Papers acknowledging the project have been prominently featured on the website within a dedicated section.
- Initiation of a series of interviews with project members to provide deeper insights into the project’s progress, featuring their expectations and challenges.
- Development of a comprehensive flyer detailing essential information about the project facilitating communication of project’s aims beyond the consortium.
- Release of the first newsletter featuring a collection of information articles aimed at providing stakeholders with valuable insights into project developments and achievements.
- Communication to the wider dementia community with publication of Pattern-Cog news items on the Alzheimer Europe website and its monthly e-newsletter.
