Reinforcement learning-based design of sequential drug treatment targeting the evolving tumour landscape with SequenTx

2026-03-04 · Source: Nature Machine Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, AI for Drug Discovery · Depth: Expert, long

Summary

SequenTx, a novel computational framework, integrates a tumor cell model with reinforcement learning to design sequential drug treatments for evolving tumor landscapes. Published in Nature Machine Intelligence on March 4, 2026, this artificial intelligence-virtual cell-inspired system considers dynamic therapy-induced transitions in tumor cellular states using transcriptome-based perturbation data. Large-scale in vitro experiments across various solid tumor types demonstrated SequenTx's effectiveness, achieving a 33% success rate (34 out of 102 combinations). In vivo studies further showed that bromodomain and extra-terminal motif inhibitor pretreatment enhanced oxaliplatin sensitivity in a melanoma xenograft model. Mechanistic analysis revealed that initial drugs induce continuous alterations in cancer cell transcriptomes, leading to enhanced responses to subsequent treatments and synergistic effects. SequenTx also identified a rationale for sequential therapy involving epigenetic inhibitors followed by other drugs, enhancing their clinical feasibility.

Key takeaway

For AI Scientists developing oncology treatments, SequenTx offers a robust framework for designing sequential drug therapies that account for tumor evolution. Your research can leverage this reinforcement learning-based approach to identify synergistic drug combinations and optimize treatment schedules, potentially improving patient outcomes by overcoming drug resistance and heterogeneity. Consider integrating similar AI-virtual cell paradigms to explore novel therapeutic strategies and enhance the clinical feasibility of existing drugs.

Key insights

SequenTx uses reinforcement learning and a virtual cell model to design effective sequential drug treatments for evolving tumors.

Principles

• Tumor cells dynamically evolve during treatment.
• Transcriptome changes can predict drug synergy.
• Epigenetic inhibitors can prime tumors for subsequent therapies.

Method

SequenTx integrates a tumor cell model with reinforcement learning, using transcriptome data to simulate dynamic cellular state transitions and predict cell viability for optimal sequential drug selection.

In practice

• Identify optimal drug sequences for specific tumor types.
• Predict synergistic drug combinations.
• Explore epigenetic drug priming strategies.

Topics

• Reinforcement Learning
• Sequential Drug Therapy
• Cancer Therapeutics
• Tumor Evolution
• Transcriptome Analysis

Code references

• DELTA-TJ/SequenTx

Best for: AI Scientist, AI Researcher, Research Scientist, Machine Learning Engineer

Related on AIssential

• Traj-Evolve: A Self-Evolving Multi-Agent System for Patient Trajectory Modeling in Lung Cancer Early Detection — Traj-Evolve combines non-parametric memory and parametric reinforcement learning for robust patient trajectory modeling in complex EHR data.
• Bridging the phenotype-target gap for molecular generation via multi-objective reinforcement learning — ExMolRL combines phenotype-guided generation with target-aware reinforcement learning for de novo drug discovery.
• AI Improving Dose Decisions and Patient Outcomes in Oncology- with Shefali Kakar of Novartis — AI-driven data integration refines oncology dose decisions, safety, and patient variability, accelerating drug development.
• 3 Questions: Building predictive models to characterize tumor progression — AI and ML tools can decode tumor evolution, especially ecDNA amplifications, to predict and counter treatment resistance.
• E-MRL: Cross-view Aligned Evidence-driven Multimodal Reinforcement Learning for Reliable 3D Tumor Analysis — Grounding Vision-Language Models in specific visual evidence via reinforcement learning significantly reduces hallucinations in medical report generation.
• Predicting and interpreting cell-type-specific drug responses in the small-data regime using inductive priors — PrePR-CT uses graph-based deep learning with cell-type-specific co-expression networks to predict drug responses in data-limited settings.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Nature Machine Intelligence.