Comparison of manual vs machine learning approaches to liver biopsy scoring for NASH and fibrosis: a post hoc analysis of the FALCON 1 study

Study Background

• While manual histological evaluation of liver biopsy tissue is the gold-standard method for fibrosis and disease staging in nonalcoholic steatohepatitis (NASH),¹ it is limited by inter- and intra-reader variability
• Machine learning models that have been trained to analyze and interpret liver
  histopathology may help improve reproducibility of NASH grading and staging²
• In liver biopsy tissue, fibrosis staging and nonalcoholic fatty liver disease activity score (NAS) results determined by PathAI, a machine learning-based approach, have been shown to correlate with those obtained from manual interpretation²
• This exploratory post hoc analysis compared manual (single central reader) and PathAI pathology scoring of liver biopsy samples from patients with NASH and stage³ fibrosis in the phase 2b FALCON 1 study