AI-Powered Insights for Personalized Prevention
- Sponsored by: TUM Startup Liv Vital
- Project lead: Dr. Ricardo Acevedo Cabra
- Scientific lead: Kamilia Zaripova, Nilo Ansari, Jonas Reinel
- TUM co-mentor: TBA
- Term: Winter semester 2025
- Application deadline: Sunday 20.07.2025
Apply to this project here

Motivation
As the world ages, healthy aging and longevity are becoming global health priorities for sustainable healthcare. In preventive medicine, physicians aim to provide personalized lifestyle advice, but current care models offer only brief, disconnected snapshots.
Meanwhile, patients use wearables, nutrition apps – yet this data remains fragmented and clinically underutilized. Liv Vital, a Munich-based digital health start-up, addresses this gap with VitalHub: a connected platform that uses AI to turn everyday health data into personalized, clinically meaningful insights.
VitalHub consists of two tools:
* A physician dashboard that integrates lifestyle and clinical data to generate prevention plans
* A patient app that mirrors key insights and facilitates ongoing communication
Together, they enable continuous, personalized prevention beyond the clinic.
The Project
The project focuses on building and testing the AI components that power the VitalHub platform. Students will work on selected AI and system modules that enable scalable, personalized prevention. Depending on interests and skillsets, possible tasks include:
* Developing LLM agents to unify and integrate diverse data sources and generating new insights through AI-driven data search.
* Automating environmental data integration into the patient's holistic profile by incorporating external risk factors.
* Generate synthetic patients using diffusion models or LLMs to safely create realistic medical data for model testing.
* Build a “what-if” simulator that shows how lifestyle changes (like reducing salt) affect patient risk.
* Predict health risks with ML models that combine vitals, labs, and rule-based logic.
* Find similar patients using embedding-based search to support clinical decisions.
* Track supplements and meds via photo-based input and dosage monitoring.
* Analyse food photos using pretrained models to estimate calories and ingredients.
* Summarize patient data into 2–3 sentence overviews using LLM prompts.
* Build an AI medical chatbot that answers doctor questions using patient data.
* Generate health plans (nutrition, activity) with LLM-based prompt tools.
Requirements:
* Experience or strong interest in AI/NLP/ML, backend/frontend development
* Ability to work with real or synthetic healthcare data in a privacy-conscious setting
* Open mindset & high motivation to collaborate with medical experts and designers
Apply to this project here