3 Questions: Building predictive models to characterize tumor progression

2026-03-10 · Source: MIT News - Artificial intelligence · Field: Health & Wellbeing — Pharmaceuticals & Biotechnology, Medical Specialties & Subspecialties, Medical Devices & Health Technology · Depth: Intermediate, short

Summary

Assistant Professor Matthew Jones at MIT is developing predictive models to characterize tumor progression, focusing on how tumors evolve to resist treatment. His research, published March 10, 2026, utilizes artificial intelligence and machine learning to decode molecular processes at genetic, epigenetic, metabolic, and microenvironmental levels. A key area of investigation is extrachromosomal DNA (ecDNA) amplifications, initially thought rare but now known to be present in about 25 percent of aggressive cancers like brain, lung, and ovarian cancers. These ecDNA amplifications accelerate tumor evolution, enabling rapid adaptation to therapies. Jones's lab aims to use computational approaches and single-cell lineage tracing technologies to understand these dynamic processes, ultimately improving patient outcomes by anticipating drug resistance and identifying new therapeutic targets.

Key takeaway

For AI Scientists developing cancer therapies, understanding the role of extrachromosomal DNA (ecDNA) amplifications is critical. Your models should incorporate ecDNA dynamics to more accurately predict tumor evolution and drug resistance. Focus on integrating single-cell lineage tracing data to pinpoint mutation timelines, which can inform strategies for intercepting tumor progression and identifying novel therapeutic targets to improve patient outcomes.

Key insights

AI and ML tools can decode tumor evolution, especially ecDNA amplifications, to predict and counter treatment resistance.

Principles

• Tumors evolve stereotypically in space and time.
• ecDNA amplifications accelerate tumor aggressiveness.
• Academic research prioritizes training future scientists.

Method

The method involves using single-cell lineage tracing technologies to study individual cell lineages and pinpoint when aggressive mutations appeared, combined with computational approaches to build predictive models.

In practice

• Stratify patients for drug response.
• Anticipate and overcome drug resistance.
• Identify new therapeutic targets.

Topics

• Tumor Progression
• Predictive Models
• Extrachromosomal DNA
• Machine Learning
• Cancer Research

Best for: AI Scientist, AI Researcher, Research Scientist, AI Student

Related on AIssential

• Understanding cancer at a genetic level with AI — AI tools like AlphaFold are democratizing advanced biomedical research in resource-limited settings.
• Reinforcement learning-based design of sequential drug treatment targeting the evolving tumour landscape with SequenTx — SequenTx uses reinforcement learning and a virtual cell model to design effective sequential drug treatments for evolving tumors.
• I Gave a Model Eyes and a Genome. Here’s What It Learned About Brain Cancer. — Cross-attention fusion of pathology and genomics significantly improves glioblastoma survival prediction over unimodal approaches.
• Computational Methods and Challenges in Cell-Free DNA Analysis for Multi-Cancer Early Detection — Multimodal ensemble approaches for cell-free DNA analysis offer the strongest promise for multi-cancer early detection clinical integration.
• Improved tumor-only variant calling and mutation burden estimation with VarNet-T — VarNet-T improves tumor-only somatic variant calling and TMB estimation using weakly supervised deep learning.
• Paper Digest: AACR 2026 Papers & Highlights — The AACR 2026 meeting features 6,000+ papers indexed by Paper Digest for rapid review.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by MIT News - Artificial intelligence.