The health status and dynamics across the lifecourse of populations is reflected in molecular, cellular, physiological, clinical and imaging data from which biomarkers can be generated using AI methodology. Such biomarkers indicate and/or predict biological age, physiological vulnerability and frailty. They can also be prognostic or generated for monitoring response to clinical and lifestyle interventions. This subtheme focuses on the methodology of generating such biomarkers , for example based on omics or brain imaging data collected in individuals, populations, patient based studies, health and socio-behavioral records and the translation of relevant biomarkers to clinical practice. There is a close collaboration with other more clinically oriented subthemes to facilitate cross-over and translation. There are two nuclei incorporated within this subtheme.
Nucleus ‘AI, classifier and biomarker development validation and implementation’
For the purpose of early recognition of vulnerability in middle to old age we develop predictive/prognostic biomarkers and composite biomarkers/diagnostic algorithms (including clocks) by applying data mining and computational tools. Biomarker profiles are composed from diverse data types: genetic/molecular/omics (DNA methylation, (epi)transcriptome/micro RNA, metabolomics, proteomics, glycomics) accelerometry/imaging (MRI/DEXA/facial images) physiological tests and challenges; in cohorts, patient and intervention studies. These biomarkers will be tested and validated in real-life clinical settings (sub-theme 1 and sub-theme 2).
Nucleus ‘Methodology for safe data sharing and integration of multi-dimensional data and fairification’
Generating and testing solutions to analyze diverse datatypes and sets jointly with many collaborating studies in AVG safe mode. Centralized datasets, personal health train (PHT). This may be on clinical health records or cohort and clinical studies based data.
Ambassadors of this sub-theme are dr. Jeroen de Bresser and prof. dr. Eline Slagboom.