Methodological Components:
1. Autonomous Workflows: Design and deployment of multi-tool AI agents Autonomous execution of analysis chains on CGM time series LLM-driven reflection and optimization of analytical decisions
2. Computer Vision: Automated recognition of food types and portion sizes from meal images using pretrained vision-language models Calorie and macronutrient estimation based on image context and metadata
3. Time-Series Analysis & Machine Learning: Detection of circadian glucose patterns (, Cosinor models, Fourier transforms, LSTM networks) Clustering and classification of metabolic response profiles across age groups and time-of-day
4. Large Language Models:
Semantic annotation and interpretation of nutrition data Automated reporting, documentation, and hypothesis formulation Contextual linking of food content, intake time, light exposure, and glucose response
Requirements:
5. Strong motivation for interdisciplinary, data-driven health research
6. Proficiency in Python; experience with OpenAI APIHuggingFace, or similar frameworks is an advantage
7. Familiarity with machine learning or computer vision methods is desirable
8. Interest in circadian biology, metabolic health, and digital prevention
9. Willingness to explore AI tools and autonomous workflows
Conditions
10. A valid certificate of study
11. For foreign students: valid residence certificate, work permit, registration certificate