PathAI and Gilead Show AI-Powered Pathology Research Models Accurately Interpret Liver Histology in Patients with NASH at AASLD 2019

Machine learning-based analyses of phase 3 trial provide quantitative assessments of liver histology and highlight heterogeneity of pathology in NASH

PathAI, a global provider of artificial intelligence (AI)-powered technology for use in pathology research, today announced the results of analyses examining the severity of nonalcoholic steatohepatitis (NASH) in liver biopsy samples from patients enrolled in the phase 3 selonsertib studies (STELLAR). Strong correlations were demonstrated between AI-powered and manual pathologist interpretations of the histological features of NASH. The data support the potential utility of machine learning models in NASH clinical trials and highlight the benefits of generating automated and quantitative assessments for staging and characterizing this disease, which affects as many as ten percent of Americans. PathAI and Gilead will present these findings at The Liver Meeting, held by the American Association for the Study of Liver Diseases from November 8-12 in Boston.

“Quantitative and reproducible assessment of liver pathology is critical for the evaluation of new therapies for NASH,” said PathAI co-founder and Chief Executive Officer Andy Beck MD, PhD. “We are thrilled to demonstrate the potential of the PathAI research platform to address this important area of drug development.”

In the first analysis (Oral Presentation #187), machine learning models were trained on liver biopsy images with more than 68,000 annotations from 75 board-certified pathologists using the PathAI research platform; images were also scored by pathologists according to the NASH Clinical Research Network (CRN) and Ishak fibrosis staging systems. The results showed that the machine learning models and the consensus of readings from the independent pathologists were in good agreement for the key histologic features of NASH. Importantly, for the staging of fibrosis, the predictions of the machine learning model were highly correlated with those of the central pathologist for both the NASH CRN (rs=0.83) and Ishak (rs=0.86) staging systems. These analyses also highlighted the potential of machine learning models to characterize fibrosis in patients with cirrhosis beyond conventional histological staging, and to do so in an automated fashion.

In the second analysis (Poster Presentation #1718), associations between the machine learning-based assessments of fibrosis, other markers of fibrosis, and the incidence of liver-related complications were evaluated in patients with cirrhosis due to NASH. The data demonstrate the heterogeneity of fibrosis in patients with cirrhosis, significant correlations between machine learning fibrosis scores and noninvasive tests of fibrosis (eg. Enhanced Liver Fibrosis [ELF] score and liver stiffness by transient elastography), and the potential to risk stratify patients with advanced fibrosis due to NASH.

“The data generated by the PathAI research platform clearly demonstrate the potential of automated, machine learning-based histologic assessment in NASH clinical trials,” said Mani Subramanian, MD PhD, Senior Vice President, Liver Diseases, Gilead Sciences. “We look forward to continued collaboration with PathAI to explore the utility of the deep learning platform to identify novel histologic features associated with treatment response and disease progression in ongoing and future trials of our NASH therapies.” Selonsertib is an investigational compound and is not approved by the U.S. Food & Drug Administration (FDA) or any other regulatory authority. Safety and efficacy have not been established.

About PathAI

PathAI is a leading provider of AI-powered research tools and services for pathology. PathAI’s platform promises substantial improvements to the accuracy of diagnosis and the efficacy of treatment of diseases like cancer, leveraging modern approaches in machine and deep learning. Based in Boston, PathAI works with leading life sciences companies and researchers to advance precision medicine. To learn more, visit