Pathology-focused AI company PathAI will present findings on the use of machine learning (ML) to enhance the scoring of nonalcoholic steatohepatitis (NASH) at the 2022 AASLD 2022 Liver Meeting.
The AASLD 2022 Liver Meeting is Nov. 4–8 in Washington, D.C., where the company will share five presentations. Boston-based PathAI developed four of them with pharmaceutical collaborators.
An oral presentation jointly developed with Gilead Sciences (Nasdaq:GILD) will touch on using exploratory analyses of NASH histology with CRN scores informed by a multi-stain ML technique. The companies will explain how a new ML-based scoring model pulling data from images stained with hematoxylin and eosin (H&E) and Masson’s Trichrome can predict NASH CRN grades and stages.
According to the presenters, the new technique has potential to reduce the high variability found in the current evaluation process of NASH biopsies. The method incorporates data from multiple hole-slide images to predict NASH CRN grades and stages on a continuous scale.
Furthermore, the continuous CRN feature scores that were extracted from these predictions were derived from the same input; as such, the PathAI model allows for a direct comparison between features associated with all four NASH histological features, revealing biologically meaningful correlations between model-derived scores and both non-invasive metrics of liver disease and gene expression patterns.
PathAI will also present two sessions in collaboration with Novo Nordisk (NYSE:NVO). The first will compare the effects of the glucagon-like peptide 1 receptor agonist semaglutide on liver histology in patients with NASH cirrhosis in an ML model assessment versus pathologist evaluation. In the second session co-presented with Novo Nordisk, PathAI will explore the association between improvement in ML-assessed steatosis area and MRI-proton density fat fraction in patients with compensated nonalcoholic steatohepatitis cirrhosis.
PathAI will join the clinical-stage biopharma 89Bio to talk about variability in liver biopsy assessment using data from a Phase 1b/2a NASH study.
“The latest research from our team aimed to explore the variabilities that exist in pathologist review of liver biopsies,” PathAI Chief Scientific Officer Dr. Mike Montalto said in a statement. “Our findings indicate that integrating artificial intelligence solutions into this process can greatly enhance the consistency and accuracy of these assessments.”
Earlier this year, we profiled Beacon Biosignals, which is using ML to tap insights from EEG data for drug development.
Filed Under: Hepatology
Tell Us What You Think!
You must be logged in to post a comment.