Multimodal AI System Demonstrates Superior Survival Prediction for Hepatocellular Carcinoma Immunotherapy
- A novel multimodal fusion (MMF) system combining CT imaging-based deep learning with clinical features achieved C-index values of 0.74 for overall survival and 0.69 for progression-free survival in predicting outcomes for unresectable hepatocellular carcinoma patients receiving immune checkpoint inhibitors.
- The MMF system outperformed traditional assessment methods by 29.8% for overall survival prediction compared to mRECIST criteria and demonstrated consistent risk stratification across diverse clinical subgroups in a multicenter validation study of 859 patients.
- Biological analysis revealed the system's predictions correlate with key molecular pathways including PI3K-Akt signaling and tumor suppressor genes like GCK, providing mechanistic insights into immunotherapy response patterns in hepatocellular carcinoma.
AbbVie
Posted 2/21/2019
Hoffmann-La Roche
Posted 12/31/2019