Related Experiment Video
Updated: Aug 22, 2025
04:14
Incorporation of a Survivable Liver Biopsy Procedure in Mice to Assess Non-alcoholic Steatohepatitis NASH Resolution
Published on: April 16, 2019
12.1K
A machine-learning approach for nonalcoholic steatohepatitis susceptibility estimation.
Fatemeh Ghadiri1,2, Abbas Ali Husseini3, Oğuzhan Öztaş4
1Department of Computer Engineering, Istanbul University Cerrahpaşa, 34320, Istanbul, Turkey. fateme.ghadiry@gmail.com.
Summary
Machine learning models accurately predict nonalcoholic steatohepatitis (NASH) risk using genetic data. This approach identifies individuals susceptible to NASH, enabling early intervention for this growing liver disease.
Area of Science:
- Genetics and Bioinformatics
- Computational Biology
- Hepatology
Background:
- Nonalcoholic steatohepatitis (NASH) is a severe liver disease with increasing global prevalence.
- Early identification of high-risk individuals is crucial for preventive and interventional strategies.
- Traditional epidemiological models have limited predictive power for NASH susceptibility.
Purpose of the Study:
- To develop a novel machine-learning approach for predicting individual NASH susceptibility.
- To utilize candidate single nucleotide polymorphisms (SNPs) and demographic data for risk prediction.
- To compare the performance of various machine-learning models for NASH prediction.
Main Methods:
- A dataset of 245 NASH patients and 120 healthy controls was analyzed.
- Genotypes for specific SNPs in CYP2E1, GCKR, and PNPLA3 genes, along with gender, were used as input features.
- Nine machine-learning models were constructed and evaluated using accuracy, precision, sensitivity, and F measure.
Main Results:
- The k-nearest neighbor (KNN) classifier, using all input features, demonstrated the highest performance.
- The top-performing KNN model achieved an 86% F measure and 79% accuracy.
- Model performance was rigorously tested on a held-out dataset.
Conclusions:
- Machine learning, leveraging genomic variations, can accurately estimate individual NASH susceptibility.
- This approach offers high accuracy, precision, and sensitivity for identifying at-risk populations.
- Genomic-based machine learning holds promise for early NASH detection and management.
