Data extraction and tumor growth modeling
Torkach explained that AI has great potential in diagnostic tasks, particularly in grading interobserver variability and aggression. Advanced algorithms give clinicians a very clear understanding of tumors from the start, and LLM extracts data from AI parameters for algorithm training, enabling effective large-scale analysis.
Multimodal models also support prediction of recurrence after discontinuation of immune checkpoint therapy in malignant melanoma using mathematical modeling of tumor growth and evolution. “Once we extract all this information from the tumor cells, we can create a model of tumor growth on the computer and see how that tumor grows,” he said. “Intratumor heterogeneity is very close to real-world data.”
profile:
Dr. Yuri Torkach He is a senior attending physician and research group leader at the Institute of Pathology, University Hospital Cologne, Germany. His “Digital and Computational Pathology” working group has research interests in diagnostic, prognostic, and predictive tools in pathology/oncology, multimodal data integration (molecular characterization), intratumoral heterogeneity and cancer evolution, large-scale language models and inference, and quality control in digital pathology.
