FDA Accepts First In Silico Drug Development Tool Under ISTAND Program to Help Predict Drug-Induced Liver Injury
On June 3, 2026, the FDA’s Center for Drug Evaluation and Research (CDER) announced a major regulatory science breakthrough by accepting the first-ever Letter of Intent (LOI) for an in silico (AI-driven) drug development tool (DDT) into its Innovative Science and Technology Approaches for New Drugs (ISTAND) DDT Qualification Program.
The Tool: DILIPredictor
The accepted DDT is an AI-Driven Digital Liver Model for Prediction of Drug-Induced Liver Injury (DILI), clinically known as DILIPredictor. Developed by Srijit Seal and collaborators (including Andreas Bender's group at the University of Cambridge and Uppsala University), the tool leverages artificial intelligence and chemoinformatics to predict human DILI during the preclinical phases of drug development.
The model works by:
- Leveraging machine learning algorithms trained on extensive chemical databases (such as the FDA's DILIst dataset of 888 compounds).
- Comparing the chemical structures of small molecule new drug candidates against historical reference drugs with known DILI risks.
- Integrating predicted in vivo and in vitro data to assess hepatotoxicity risk before entering Phase I clinical trials.
Regulatory and Scientific Impact
Drug-Induced Liver Injury (DILI) is one of the most severe safety concerns in pharmaceutical development, representing a leading cause of clinical trial terminations and post-market drug withdrawals. Current animal and in vitro assays often fail to accurately replicate human liver toxicity.
By accepting DILIPredictor, the FDA aims to:
- Minimize Animal Testing: Aligning with the FDA's commitment to New Approach Methodologies (NAMs) and the "3Rs" (replacement, reduction, and refinement) of animal testing in drug development.
- Expedite Drug Discovery: Providing a validated "weight-of-evidence" screening tool that allows developers to de-risk compounds earlier in the pipeline.
- Establish a Three-Step Pathway: This LOI acceptance marks the first of a three-step qualification process. The developers will next submit a formal Qualification Plan, followed by a full qualification package. Once fully qualified, any pharmaceutical sponsor can use the tool in regulatory filings without needing to re-validate its performance.
"With this action, FDA is supporting innovative ways to optimize the development and evaluation of potential new therapies... The AI-Driven Digital Liver Model shows promise in assessing the risk of hepatotoxicity during preclinical phases of drug development." — Michael Davis, MD, PhD, Acting CDER Director, and Jeffrey Siegel, MD, CDER Office of New Drugs, June 3, 2026
"DILI Predictor is made from a joint effort from Andreas Bender's group at the University of Cambridge... Overall, the DILIPredictor model improves the detection of compounds causing DILI with an improved differentiation between animal and human..." — Srijit Seal, University of Cambridge / Uppsala University, June 2026