Lecture | Webinar
AI & Data Science @ Faculty of Social and Behavioural Sciences
- Friday 29 May 2020
- Kaltura Live Room opens for the public at 12:50.
- Kaltura Live Room
The Data Science Research Programme, SAILS and the Faculty of Social and Behavioural Sciences have joined forces in this unique event at the intersection between these three areas of science and the possibilities that arise from them.
In a series of three webinars, you will get insight into multi-view learning in biomedical data, imaging brain networks and innovative approaches on parenting from a family perspective.
The webinars are accessible for free and open to everyone.
Join the webinar via Kaltura Live Room
Kaltura Live Room works best in Edge, Chorme and Firefox. Make sure you activate your camera and microphone beforehand in order to interact with the speaker and participate in discussion. The room opens for the public at 12:50.Enter Kaltura Live Room
|Friday 29 May 2020 13:00||
Serge Rombouts - Professor of Methods of Cognitive Neuroimaging
Imaging Brain Networks: pharmacological manipulation and individual prediction of cognitive decline.
|Friday 5 June 2020 13:00||
Wouter van Loon - PhD Candidate
Selecting views in multi-view learning.
|Friday 12 June 2020 13:00||
Bernet Elzinga - Professor of Stress-related Psychopathology
Innovative approaches on parenting from a family perspective.
Serge Rombouts - Imaging Brain Networks:
pharmacological manipulation and individual prediction of cognitive decline.
Individual dementia biomarkers are strongly needed for i) early (differential) diagnosis in the stages preceding significant cognitive loss, ii) marking disease stage and monitoring disease course and possible treatment effects, and iii) guidance for drug development targeting the early phase of the disease. The discovery of an early biomarker is considered the holy grail in dementia research.
With MRI, both brain structure, anatomical connections and functional brain networks can be studied. Different (early) stages in dementia and different dementia types can each be characterized by a combination of MRI-derived features, contained in a ‘multivariate multidimensional MRI biomarker’.
We are studying the combination of MRI features with maximal classification accuracy of individual patients according to disease severity and dementia type. Additionally, we are evaluating the sensitivity of this biomarker for early diagnosis and prediction of future decline rate using longitudinal studies in
1. individuals with cognitive impairment
2. asymptomatic mutation carriers, destined to develop frontotemporal lobe dementia
3. a large middle-aged and elderly population of non-demented individuals that is followed for incidence of Alzheimer’s disease
We study whether these innovations will lead to an early individual MRI biomarker that contributes to the quality of early dementia diagnosis, is useful for monitoring of disease course and treatment, and will provide new leads for drug development studies.
Parallel to this, we are developing methods of pharmacological FMRI, to study effects of various pharmacological manipulations on functional brain connectivity in young, elderly and in dementia patients.
Wouter van Loon - Selecting views in multi-view learning
Integrating information from different feature sets describing the same set of objects or persons is known as multi-view learning. In biomedical research, such feature sets (views) may correspond to different data sources such as medical imaging modalities, questionnaires, and omics data. Views can also be defined within data obtained from the same source, for example as different feature sets derived from the same image, as different brain regions, or as gene sets. Integrating the information from different views can increase the accuracy of medical classification models. However, collection of biomedical data can be expensive and/or burdening for patients. Identifying the views that are most important for prediction can improve the understanding of disease and can contribute to reducing the amount of required data collection. This leads to a group-wise feature selection or ‘view selection’ problem. In this talk I will discuss this view selection problem and a method we have been developing to tackle it.
Bernet Elzinga - Innovative approaches on parenting from a family perspective
Sensitive parenting, including parental warmth, is crucial for later mental well-being. Child maltreatment, particularly physical and emotional abuse and neglect, often occurs within the family context. Studying on the impact of specific parenting styles on adult mental well-being can be methodologically challenging, since families consist of multiple family members (i.e., parents, children, siblings) with each their own perspective. So far, commonly used methods focus usually on the perspective of one single individual and investigate the impact of individual parent-to-child interactions, while the complex interactions and the different perspectives of family members are not taken into account. To further contribute to the ecological validity of the research on parenting and child maltreatment, incorporating the perspectives of different individuals living in the same household is important. In this lecture, I will present a number of studies on parenting and child maltreatment, using a variety of research methods, e.g. observations of interactions, fMRI research and multi-informant assessments, while integrating different family perspectives (i.e. parents versus children, brothers and sisters) and discuss its impact on individual’s and family’s lives. These approaches may further advance our knowledge relevant for assessment, prevention and (family) interventions.