Non-alcoholic steatohepatitis (NASH) disease severity is typically assessed via ordinal scoring of tissue biopsies. However, ordinal systems lack the sensitivity required to detect the heterogeneity that exists in liver tissue or capture subtle changes that may indicate improvement or worsening of disease. Additionally, pathologist scoring of NASH histologic features (steatosis, ballooning, lobular inflammation, and fibrosis) is difficult, error-prone, and subject to intra- and inter-reader variability.
Here, we report machine learning-facilitated continuous scoring of histologic features by PathAI’s AI-based Measurement of NASH Histology (AIM-NASH) algorithms in a retrospective analysis of a datasets from two completed NASH clinical trials. The prognostic utility of continuous scores was evaluated and compared against ordinal scoring for predicting patient outcomes Iyer et al.
Here, we report machine learning-facilitated continuous scoring of histologic features by PathAI’s AI-based Measurement of NASH Histology (AIM-NASH) algorithms in a retrospective analysis of a datasets from two completed NASH clinical trials. The prognostic utility of continuous scores was evaluated and compared against ordinal scoring for predicting patient outcomes Iyer et al.
