Artificial intelligence improves prediction of cancer drug resistance

A comprehensive review recently published in Current Molecular Pharmacology (2026, Volume 19, Pages 85–96) examines the rapidly evolving landscape of computational tools for predicting tumor drug resistance. Led by Jia Wang, Hong‑Rui Zhu, and corresponding authors Zhi‑Chun Gu and Hou‑Wen Lin from Shanghai Jiao Tong University School of Medicine, the article systematically maps how artificial intelligence-particularly machine and deep learning-is being harnessed to integrate multi‑omics data from large‑scale repositories such as TCGA and GDSC. These approaches are helping to decode resistance mechanisms across chemotherapy, targeted therapy, and immunotherapy, while also pointing to novel predictive dimensions such as cancer‑associated thrombosis.

The authors emphasize that standardized databases and sophisticated preprocessing pipelines are now essential for transforming heterogeneous genomic, transcriptomic, and clinical data into reliable model inputs. Yet they caution that data sparsity, batch effects, and the "black‑box" nature of many deep‑learning models remain substantial barriers to clinical adoption. "The inherent trade‑off between model accuracy and interpretability undermines clinician trust and limits real‑world adoption," notes Dr. Gu. To address this, the review advocates for explainable AI frameworks, multimodal fusion strategies, and the integration of dynamic liquid‑biopsy monitoring to capture resistance evolution in real time.

Looking forward, the team calls for a paradigm shift towards specialized tools for high‑risk subgroups, particularly patients with cancer‑associated thrombosis. By incorporating coagulation‑related signatures and longitudinal thrombotic markers, these next‑generation models could offer actionable predictions that guide combined anticancer and anticoagulant therapies. The authors also urge the establishment of unified data standards, prospective clinical validation, and interdisciplinary collaboration to bridge the gap between computational innovation and bedside application. "Our goal is to move beyond generic predictions and deliver tailored insights for the patients who need them most," adds Professor Lin. The review concludes that with sustained efforts in data integration, interpretability, and clinical translation, AI‑driven resistance prediction holds transformative potential for precision oncology.