Multi-scAle high-order neTwork modelling in the stUdy of human aging: A data-dRiven perspectIve using Topology and geometrY

Projektbeschreibung

Mit Neuroimaging Alterungsprozesse kartieren

Das Altern und die damit verbundenen Krankheiten zu verstehen, bildet eine der komplexesten Herausforderungen der modernen Biologie. Mit modernsten neurobildgebenden Verfahren wie der funktionellen Magnetresonanztomografie und Elektroenzephalografie kann die Hirnaktivität gemessen werden. Die Entwicklung präziser Diagnostikwerkzeuge gestaltet sich jedoch aufgrund von Rauschen und Datenqualitätsproblemen schwierig. Obwohl die topologische und geometrische Datenanalyse vielversprechend erscheint, behindern Einschränkungen der Algorithmen den Fortschritt. Im Rahmen des innerhalb der Marie-Skłodowska-Curie-Maßnahmen unterstützten Projekts MATURITY werden Zeitreihendaten aus Datenbanken über das Altern mithilfe eines mehrschichtigen Netzwerkansatzes höherer Ordnung analysiert. Das Projektteam verfügt über einen einzigartigen Zugang zu Daten aus allen Lebensphasen und auf verschiedenen Ebenen, darunter Multiomik-, Exposom-, Mikrobiom- und Neuroimagingdaten sowie klinische Daten von Tausenden Patientinnen und Patienten. Insgesamt wird geklärt, wie der Alterungsprozess über biologische Prozesse abläuft.

Ziel

A full understanding of ageing is still elusive, as well as the consequences in terms of age-related diseases. Several approaches have been proposed to overcome the challenges of cure and early diagnosis of brain diseases, such as Parkinson's and Alzheimer's. With the advent of technology, different approaches are used in order to measure brain activity, such as fMRI, PET scans, EEG/MEG scans, to name a few. However, in order to propose accurate tools for diagnosis, the choice of an appropriate data measurement that can be noise-free and endowed with the correct features is far from being an easy task. Topological and geometrical data analysis emerged in the last decade as a very promising tool. Yet, the lack of efficient approaches, as well as computational algorithms, prevents rapid advances in the field of big data. These limitations also prevent the analyses from going beyond low-dimensional data inspection, also limiting the information provided. The present project aims to propose systematic answers to some of these issues, with the goal of analysing and classifying (both topologically and geometrically) time-series from ageing databases in a multi-layer higher-order network. To this end, the RF has unique access to human lifespan (20 to over 100 years of age) multi-scale data (from multi-omics, exposome, microbiome, neuro-imaging to phenotypic) from the NIH/USA (National Institute on Aging) and multi-modal clinical data of thousands of patients. Being an expert in multivariate time series to generate higher-order information from, he will develop an efficient, optimized geometrical/topological tool for computing higher-order invariants. The RF hypothesizes a relation between the long genes and the susceptibility of the human brain to age-related disease. He anticipates that merging multi-omics and brain data, together with the proposed multi-layered topological workflow classification, will allow him to generate a proof-of-concept model for ageing pathways.