.png){kind=icon}
.png){kind=icon}
Return
Integrating in vitro and in silico NAMs for enhanced prediction of drug-induced liver injury
.png){kind=icon}
OA
AI
T
S
Y
W
DOI：10.1016/j.drudis.2026.104681.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Drug-induced liver injury remains a persistent challenge in drug development and regulatory review, driving active investigation of new approach methodologies to improve prediction. • This study presents a Confidence-Guided Integration rule to combine an in silico new approach methodology model (DeepDILI) with in vitro new approach methodologies (six human liver spheroid assays) to enhance drug-induced liver injury prediction. • Confidence-Guided Integration improved predictive accuracy by 9–55% compared with human liver spheroid assays alone in five of six datasets. It also identified a high-confidence subset of drugs with concordant predictions, achieving 86–100% accuracy. The most significant gain with Confidence-Guided Integration was for discordant cases, with an error reduction of up to 53%. • Confidence-Guided Integration produced drug-induced liver injury classifications comparable to a published Liver-Chip system (Emulate Liver-Chip) in an eight-drug evaluation. • This strategy demonstrates a practical new approach methodology integration (such as in silico and in vitro), supporting the 3Rs (Replacement, Reduction and Refinement) and the FDA’s 2025 road map to reduce animal testing.
Keywords:
NAMs
3Rs
DILI
Spheroid Assay
AI
In Vitro
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
7.5
Papers:
6.3K
Citations:
2.3W
Organization
No organization information available
Cited Papers