Innovative Diagnostics, Modelling and Digital Hepatology - Assessment of portal hypertension severity using machine learning models in patients with compensated cirrhosis

Aug 5, 2026

Upload by Finn

In individuals with compensated advanced chronic liver disease, portal hypertension severity predicts risk of decompensation. The study evaluates machine learning models built from widely available laboratory parameters to estimate portal hypertension severity, targeting clinically significant portal hypertension (HVPG ≥10 mmHg) and severe portal hypertension (HVPG ≥16 mmHg). Models are trained on an internal Vienna cohort, optimized on a merged cohort, and tested on external multicenter data to assess prediction robustness and generalizability, enabling a practical online risk-stratification tool.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Background & Aims

Methods

Results

Conclusions

Research Article Innovative Diagnostics, Modelling and Digital Hepatology Assessment of portal hypertension severity using machine learning models in patients with compensated cirrhosis Authors Jií Reini , Oleksandr Petrenko, Benedikt Simbrunner, , Stefan Kubicek, Mattias Mandorfer, Thomas Reiberger Correspondence [thomas.reiberger@meduniwien.ac.at](thomas.reiberger@meduniwien.ac.at) (T. Reiberger). Graphical abstract Development of machine learning models 1 2 3 Internal cohort of patients (VICIS, NCT03267615) with compensated advanced chronic liver disease (cACLD; n = 163) Diagnostic workup including a wide set of laboratory parameters, HVPG and liver stiffness measurements Imputation of missing data Missing data Imputed data Validation on external cohort 4 Feature selection and evaluation of ML models to predict CSPH and HVPG ≥16 mmHg 5 6 7 8 Multicenter validation cohort (n = 1,232 cACLD patients) Data verification and pre-processing Evaluation of threeand five-parameter models Split-traning and evaluation on the merged cohort to predict CSPH and HVPG ≥16 mmHg Highlights 􀀁 Models that can non-invasively assess portal hypertension severity are an unmet clinical need. 􀀁 Machine learning models trained on 3/5 laboratory parameters enabled non-invasive assessment of portal hypertension severity. 􀀁 These models could predict portal pressures of 10 mmHg or 16 mmHg in individuals with compensated cirrhosis. 􀀁 An online tool based on these models has been made available and can be used for portal hypertension risk stratiﬁcation. Impact and implications We used machine learning models based on widely available laboratory parameters to develop a non-invasive model to predict the severity of portal hypertension in individuals with compensated cirrhosis, who currently require invasive measurement of hepatic venous pressure gradient. We validated our ﬁndings in a large multicentre cohort of individuals with advanced chronic liver disease (cACLD) of any cause. Finally, we provide a readily available online calculator, based on 3 (platelet count, bilirubin, international normalised ratio) or 5 (platelet count, bilirubin, activated partial thromboplastin time, gamma-glutamyltransferase, choline-esterase) widely available laboratory parameters, that clinicians can use to predict the likelihood of their patients with cACLD having clinically signiﬁcant or severe portal hypertension. [https://doi.org/10.1016/j.jhep.2022.09.012](https://doi.org/10.1016/j.jhep.2022.09.012) © 2022 The Authors. Published by Elsevier B.V. on behalf of European Association for the Study of the Liver. This is an open access article under the CC BY license ([http://creativecommons.org/licenses/by/4.0/](http://creativecommons.org/licenses/by/4.0/)). J. Hepatol. 2023, 78, 390–400 Research Article Innovative Diagnostics, Modelling and Digital Hepatology Assessment of portal hypertension severity using machine learning models in patients with compensated cirrhosis Jií Reini1 ,†, Oleksandr Petrenko1 ,2 ,3 ,4 ,†, Benedikt Simbrunner1 ,2 ,3 ,4, Benedikt S. Hofer1 ,2 ,3 ,4, Filippo Schepis5, Marco Scoppettuolo5 , Dario Saltini5, Federica Indulti5, Tomas Guasconi5, Agustin Albillos6, Luis Téllez6, Càndid Villanueva7, Anna Brujats7, Juan Carlos Garcia-Pagan8 , Valeria Perez-Campuzano8, Virginia Hernández-Gea8, Pierre-Emmanuel Rautou9, Lucile Moga9, Thomas Vanwolleghem10 , 11, Wilhelmus J. Kwanten10 , 11, Sven Francque10 , 11, Jonel Trebicka12 , 13 , 14, Wenyi Gu12 , 13, Philip G. Ferstl12 , 13, Lise Lotte Gluud15, Flemming Bendtsen15 , Søren Møller16, Stefan Kubicek1, Mattias Mandorfer3 ,4, Thomas Reiberger1 ,2 ,3 ,4 ,* Journal of Hepatology 2023. vol. 78 j 390–400 Background & Aims: In individuals with compensated advanced chronic liver disease (cACLD), the severity of portal hypertension (PH) determines the risk of decompensation. Invasive measurement of the hepatic venous pressure gradient (HVPG) i

FAQs

What problem does the study address in compensated cirrhosis?

It addresses how to assess the severity of portal hypertension, which determines the risk of decompensation, without relying solely on invasive testing.

Which inputs do the machine learning models use?

The models are based on 3 or 5 routinely available laboratory parameters, including platelet count, bilirubin, international normalised ratio, and additional coagulation and liver-related markers for the 5-parameter version.

How were the models evaluated and what is the key limitation?

Performance was tested in external validation datasets after internal training and merged-cohort optimization; results were reliable internally but heterogeneous across external datasets.

Aug 5, 2026

Upload by Finn