What question did this study set out to answer?

To create a predictive model for objective response and progression-free survival in patients with HBV-related unresectable HCC.

July 12, 2026Scientific ReportsOpen Access

A deep learning model to predict objective response to TACE and TKI-based therapy in HBV-related uHCC

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial demonstrates a deep learning model predicts treatment response in uHCC, suggesting tailored therapies are feasible.

To create a predictive model for objective response and progression-free survival in patients with HBV-related unresectable HCC.
Retrospective analysis of 243 patients, divided into training (n=168) and test datasets (n=75).
Developed three models: Clinical Model (C-Model), Machine Learning Radiomics Model (ML-Model) using 1,479 CT features, and Deep Learning Model (DL-Model) based on ResNet-50.
Evaluated model performance using AUC and Kaplan-Meier analysis.
The DL-Model achieved an AUC of 0.851 (95% CI: 0.747–0.954), outperforming the C-Model (AUC = 0.586, P < 0.05) and ML-Model (AUC = 0.709).
Predicted responders using the DL-Model had significantly prolonged progression-free survival (P = 0.011).
Grad-CAM analysis showed responders had centralized tumor activation, while non-responders exhibited multifocal activation patterns.

Ding et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark