FDA Opens Door to AI Scoring in Drug Trials

FDA clears PathAI tool for MASH studies, boosting the push for reproducible digital pathology

10 Dec 2025

The US Food and Drug Administration has expanded the role of artificial intelligence in drug development by qualifying AIM MASH AI Assist, PathAI’s system for analysing liver biopsy slides in studies of metabolic dysfunction. The approval applies only to clinical research rather than routine diagnosis but indicates a change in how regulators view machine learning in trial settings.

Developers of treatments for MASH have long faced delays linked to manual biopsy review, where expert opinions often diverge. PathAI’s tool processes digital slides and produces more consistent scores, easing a bottleneck that has slowed large trials. Early data reviewed by the FDA suggest the model performs at a level similar to experienced pathologists, with steadier results across different sites.

Analysts say the decision moves the sector towards digital standards in study design. Andy Beck, PathAI’s chief executive, said the technology “brings a level of uniformity global trials rarely achieve”, reducing the burden on specialist reviewers and improving data quality. Reuters reported that sponsors can now integrate AI-generated readings into protocols, a shift that could shorten timelines for a range of drug candidates.

Interest in digital pathology is rising among investors, research networks and health systems, which see AI-based imaging as part of core research infrastructure. While AIM MASH AI Assist is not tied to a corporate transaction, its qualified status is influencing partnership plans among developers seeking digital capacity. The trend points to emerging ecosystems that link imaging, analytics and workflow tools.

Questions remain over data security, patient privacy and the safeguards needed as algorithms take on sensitive clinical tasks. Specialists note that transparency around model performance will be important as more systems seek regulatory backing.

The FDA’s decision suggests a move towards faster and more data-rich research environments, with many industry leaders viewing it as an early stage in a wider expansion of AI-enabled drug development.