FDA Qualifies First AI Drug Development Tool, Will Support Histology Assessment in MASH Trials

The FDA has qualified the first AI drug development tool called the AI-Based Histologic Measurement of NASH (AIM-NASH), for use in clinical trials assessing therapies for metabolic dysfunction-associated steatohepatitis (MASH). The tool helps pathologists score liver biopsy samples and may help streamline histologic assessments in studies of new MASH treatments.

MASH (previously known as NASH or nonalcoholic steatohepatitis) is a progressive form of fatty liver disease marked by fat buildup, inflammation and fibrosis.

A recent modeling study estimated that about 5.8% of US adults, or roughly 15 million people, had MASH in 2020. This included about 6.7 million individuals with clinically significant fibrosis (stage F2 or higher). These figures are expected to rise steadily through 2050.

MASH is closely linked to obesity and type 2 diabetes and is often diagnosed late, sometimes only after complications such as cirrhosis or liver cancer have developed.

Underdiagnosis is still a major barrier to earlier care. Many people with MASH are not identified until they present with advanced liver disease. Policy experts have noted that AI may help improve liver risk assessment using routine data from EHRs, supporting broader efforts to close the diagnostic gap.

The AIM-NASH system is a cloud-based AI tool that analyzes digital images of liver tissue and provides standardized scores for steatosis, hepatocellular ballooning, lobular inflammation and fibrosis. These features are routinely assessed in MASH trials and align with the NASH Clinical Research Network scoring system.

The qualification gives sponsors an FDA-recognized way to use AI for histologic scoring, but it does not replace the need for expert pathology review or change biopsy-based endpoints in ongoing trials. The agency emphasized that human oversight remains essential, with AI intended to function strictly as an assistive tool.

FDA qualification relied on validation studies showing that AIM-NASH-assisted evaluations matched well with expert consensus and performed similarly to individual pathologists.

The agency noted that the tool could help reduce variability in biopsy scoring, which has contributed to variability in study outcomes in MASH programs over the past decade.

XTALKS WEBINAR: Mastering Inspection Readiness for FDA’s AI Tool Elsa

Live and On-Demand: Monday, December 15, 2025, at 1pm EST (10am PST)

Register for this free webinar to learn how inspection readiness can evolve to meet the demands of AI-enabled regulatory oversight.

This decision follows a year of increased FDA activity in AI oversight, including the rollout of Agentic AI, enhancements to the TEMPO real-world evidence program for chronic disease tools and updates to draft guidance on AI- and machine learning-enabled medical devices.

AI applications are also growing in MASH research. In 2025, HistoIndex launched FibroSIGHT Plus, an AI-enabled fibrosis assessment service that uses stain-free imaging and quantitative analysis to support more consistent biopsy evaluation in research and clinical settings.

Non-invasive methods are also being developed. In November 2025, Hepta released early clinical data showing that its AI-powered cfDNA epigenetics assay detected MASH with significant fibrosis, achieving a diagnostic area under the curve (AUC) of 0.86 (a measure of how accurately a test distinguishes between disease and no disease) and generating fewer false positives than standard blood tests. This shows the potential for liquid biopsy tools to support earlier identification and monitoring.

A 2025 review of AI in metabolic dysfunction-associated steatotic liver disease (MASLD)  noted that the use of digital pathology, imaging-based fibrosis measurement and clinical data-driven prediction models is expanding, though most applications are still in early stages of clinical adoption.

Researchers have also developed AI models that estimate fibrosis using routine blood and clinical data. In one 2025 study, the model identified patients with significant fibrosis with accuracy comparable to established non-invasive tests, without relying on platelet counts or specialized imaging techniques such as elastography, which are not always available in routine care.

Next steps for the AIM-NASH tool were not disclosed in the FDA news, but qualification allows its use across MASH drug development programs without requiring individual sponsors to resubmit validation packages.

If you want your company to be featured on Xtalks.com, please email [email protected].