A recent comprehensive review published in Current Molecular Pharmacology examines advances in computational tools for predicting tumor drug resistance, according to a June 26 article. The study, 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, maps the use of artificial intelligence—particularly machine and deep learning—to integrate multi-omics data from large-scale repositories such as TCGA and GDSC.
The review describes how these approaches are being used to decode mechanisms of resistance across chemotherapy, targeted therapy, and immunotherapy. It also highlights emerging predictive areas such as cancer-associated thrombosis. The authors say that standardized databases and advanced preprocessing pipelines are now essential for transforming heterogeneous genomic, transcriptomic, and clinical data into reliable model inputs.
However, the researchers caution that issues like data sparsity, batch effects, and the opaque nature of many deep-learning models remain significant barriers to clinical adoption. "The inherent trade-off between model accuracy and interpretability undermines clinician trust and limits real-world adoption," said Dr. Gu. To address these challenges, the review recommends explainable AI frameworks, multimodal fusion strategies, and integrating dynamic liquid-biopsy monitoring to track resistance evolution in real time.
Looking ahead, the team calls for specialized tools targeting high-risk subgroups such as patients with cancer-associated thrombosis. By including coagulation-related signatures and longitudinal thrombotic markers in future models, they suggest actionable predictions could be developed to guide combined anticancer and anticoagulant therapies. The authors also urge unified data standards, prospective clinical validation studies, and interdisciplinary collaboration to help translate computational advances into clinical practice.
"Our goal is to move beyond generic predictions and deliver tailored insights for the patients who need them most," said Professor Lin. The review concludes that ongoing efforts in data integration, interpretability improvements, and clinical translation could enable AI-driven resistance prediction to play a transformative role in precision oncology.
