Cirrhosis detection in Hepatitis C: Extra Trees hits 96.92%

An Extra Trees ensemble detected cirrhosis in Hepatitis C patients with 96.92% accuracy, in a paper submitted to arXiv on 25 June 2026. The study used a dataset of 2,038 Egyptian patients from the UCI Machine Learning Repository and trained four algorithms: Random Forest, Gradient Boosting Machine, Extreme Gradient Boosting, and Extra Trees.

What did the study do and find?

The paper trained four ensemble machine learning models on a 2,038-patient Hepatitis C dataset and found that an Extra Trees model performed best, achieving an accuracy of 96.92%, a recall of 94.00%, a precision of 99.81%, and an area under the ROC curve of 96% while using 16 of the 28 available features. The dataset and those model choices are described in the abstract; the authors frame the work as detecting cirrhosis earlier to avoid complications of liver disease.

The study authors are Abrar Alotaibi and seven co-authors. They obtained the dataset of 28 attributes for 2,038 Egyptian patients from the University of California at Irvine Machine Learning Repository. The paper presents the models as "explainable ensemble-based machine learning models" for diagnosing cirrhosis in Hepatitis C patients; the Extra Trees classifier required only 16 features to reach the reported performance metrics.

How did the models compare?

Only the Extra Trees model's numeric performance appears in the abstract: it outperformed the other three models on the reported metrics. The authors state they trained a Random Forest, a Gradient Boosting Machine, an Extreme Gradient Boosting model, and an Extra Trees model, but the abstract gives detailed metrics only for the Extra Trees classifier. The Extra Trees run delivered 96.92% accuracy, 94.00% recall, 99.81% precision, and a 96% area under the receiver operating characteristic curve.

The paper emphasizes feature reduction: of the 28 original attributes, the Extra Trees model used 16 features to reach the reported results. The abstract does not supply numeric comparisons for the Random Forest, Gradient Boosting Machine, or Extreme Gradient Boosting runs, only noting that four algorithms were trained on the same dataset.

Why it matters

Cirrhosis often develops over years with few symptoms, and earlier detection reduces progression to liver failure and related complications. The paper points out that, despite machine learning's use in diagnosing other diseases, no prior studies had used ML specifically to detect cirrhosis in patients with Hepatitis C. A high-precision, high-recall classifier that uses a subset of routine features could shorten the path to earlier intervention if its results generalize beyond the UCI cohort.

What to watch

The immediate signal to follow is whether these models are validated on external clinical cohorts beyond the 2,038-patient UCI dataset. Replication on independent datasets or prospective clinical validation would confirm whether the Extra Trees model’s 96.92% accuracy and 99.81% precision hold up in different populations.

Bibliographic note: the paper was submitted to arXiv on 25 June 2026 and lists a journal reference of Computation 2023, 11, 104 in its metadata.