The biological aging of human organ systems represents the interplay of age and chronic diseases with lifestyle factors. We can establish models trained on a healthy cohort to simulate healthy aging and then identify the subjects with accelerated (predicted age > chronological age) and decelerated (predicted age < chronological age). Tian et al., 2023 and Wen et al., 2024 define the aging of different organ systems utilizing non-imaging biomarkers. We want to add imaging biomarkers to the already existing non-imaging biomarkers and improve the accuracy of age prediction. We aim to identify new associations between accelerated (predicted healthy age > chronological age) aging and chronic diseases and lifestyle factors.

1. Literature search for aging with imaging data
2. Developing imaging and multi-modal models for organ-wise age predictions (raw MRI or/and extracted radiomics)
3. Longitudinal analysis of subjects (examples below and can be extended)
   1. CVD vs. Healthy cohort: Do they have accelerated age, and in which organ systems?
   2. Timepoint 1 and timepoint 2 of healthy subjects. What is the statistical difference, and what organs age faster?
   3. Identifying normal → accelerated or accelerated → normal or decelerated → accelerated. What are the lifestyle differences?

1. Close supervision, weekly meetings, access to state-of-the-art computer hardware
2. Exciting state-of-the-art research project with many possibilities to bring in your own ideas
3. If required, medical feedback from doctors

1. Advanced programming skills in Python and common DL and ML frameworks (Pytorch and scikit-learn)
2. Strong foundation in statistics and deep learning methodologies
3. Independent working style with a strong interest in teamwork and methodic research

Please send your CV and transcript using the subject “MSc Thesis: Multi-modal Longitudinal Heterogeneous Aging in UK Biobank” to:

Dmitrii: dmitrii.seletkov@tum.de

Sophie: sophie.starck@tum.de

Please also include brief summaries of your prior deep learning projects (including your contributions to the project and the used framework) and, if available, provide links to the codebases (e.g., your GitHub profile).

References: 

1. Tian et al., 2023
2. Wen et al., 2024
3. Chen et al., 2023