BEGIN:VCALENDAR VERSION:2.0 PRODID:-//Talks.cam//talks.cam.ac.uk// X-WR-CALNAME:Talks.cam BEGIN:VEVENT SUMMARY:Advances in building patient-specific agent-based models in cancer - Paul Macklin (Indiana University Bloomington) DTSTART:20250918T131500Z DTEND:20250918T140000Z UID:TALK235195@talks.cam.ac.uk DESCRIPTION:\n\n\n\n\n\nTumors are a mixture of malignant cancer cells\, s tromal cells\, and immune cells that interact as dynamically evolving ecos ystems. Over the last ten years\, there has been substantial progress in b uilding patient-tailored simulation models of cancer: a key component of d igital twins. In the context of cancer immunology\, single-cell effects an d pairwise cell-cell interactions are paramount. Agent-based models (ABMs) \, which simulate individual cells as software agents that interact in sim ulated tissue environments\, are well-suited to simulating these cancer-im mune systems\, but they generally require substantial hand-coding (in C++\ , Python\, etc.). Beyond creating a substantial barrier to entry for domai n experts and multidisciplinary teams\, this has also hampered model reuse \, extensibility\, maintenance\, and reproducibility. In this talk\, we pr esent a new conceptual framework&mdash\;a cell behavior hypothesis grammar &mdash\;that uses natural language statements (cell rules) to create mathe matical and simulation models in real time without writing code. This enab les systematic integration of biological knowledge and multi-omics data to generate \;in silico \;models with user-friendly online tools\, e nabling virtual &ldquo\;thought experiments&rdquo\; that interactively tes t and expand our understanding of multicellular systems and generate new t estable hypotheses. We demonstrate its use in developing both \;de nov o mechanistic models and those informed (and initialized) by multi-omics d ata\, with focused examples on models of hypoxic breast cancer\, tumor-imm une interactions\, and virtual combination immunotherapy trials in pancrea tic ductal carcinoma (PDAC). We also show how rapid model creation (via th e grammar) can be combined with high-performance computing (HPC) to accele rate discovery in cancer biology and development of reusable components fo r digital twins. We close with a discussion of how language-focused modeli ng will open new doors to combine human and machine intelligence in mathem atical oncology. \;\nThese results--recently published in Cell--are th e joint work of a broad coalition including Elana Fertig (University of Ma ryland)\; Laura Heiser\, Young Hwan Chang\, Lisa Coussens\, and Joe Gray ( Oregon Health and Sciences University)\, Genevieve Stein-O'Brien (Johns Ho pkins)\, and others with my lab at Indiana University. \;\n\n\n\n\n\n LOCATION:Seminar Room 1\, Newton Institute END:VEVENT END:VCALENDAR