Malignant tumors do not act in local isolation. They alter the entire organism by affecting metabolism, skeletal muscle, adipose tissue, and internal organs. These systemic changes, in turn, influence the course of disease, for example by affecting treatment tolerance or promoting the development of new metastases. This interplay between the tumor and the host organism is referred to as the tumor macroenvironment.

Imaging studies have so far largely focused on individual organs at a single point in time. How organs and tissues interact with one another, and how this network changes over the course of disease, has not yet been systematically characterized.

This project aims to address this gap by developing an AI model based on routine CT scans from patients with gastrointestinal tumors. Organs and tissues will be converted into quantitative features, and their interactions will be modeled using a time-sensitive algorithm. The goal is to predict overall survival, patterns of metastasis, and treatment toxicity, thereby enabling more precise risk stratification.