Author information

¹Newcastle University, Newcastle upon Tyne, United Kingdom.

²Novartis Institute for Biomedical Research, Cambridge, Massachusetts, United States of America.

³Novartis Pharmaceuticals, East Hanover, New Jersey, United States of America.

⁴Institute of Cardiometabolism and Nutrition, Paris, France.

⁵Department of Medicine II, University Medical Center Homburg and Saarland University, Homburg, Germany.

⁶University of Torino, Turin, Italy.

⁷University Hospital Würzburg, Würzburg, Germany.

⁸Servicio Andaluz de Salud, Seville, Spain.

⁹University of Bern, Bern, Switzerland.

¹⁰Linköping University, Linköping, Sweden.

¹¹Antwerp University Hospital, Antwerp, Belgium.

¹²University of Helsinki, Helsinki, Finland.

¹³University of Cambridge, Cambridge, United Kingdom.

¹⁴Università degli Studi di Milano, Milan, Italy.

¹⁵Università Cattolica del Sacro Cuore, Rome, Italy.

¹⁶University of Oxford, Oxford, United Kingdom.

¹⁷Medical School of National & Kapodistrian University of Athens, Athens, Greece.

¹⁸AMC Amsterdam, Amsterdam, The Netherlands.

¹⁹Translational & Clinical Research Institute, Faculty of Medical Sciences, Newcastle University, Newcastle upon Tyne, United Kingdom.

²⁰Newcastle NIHR Biomedical Research Centre NUTH NHS Trust, Newcastle upon Tyne, United Kingdom.

Abstract

Aims: Metabolic dysfunction Associated Steatotic Liver Disease (MASLD) outcomes such as MASH (metabolic dysfunction associated steatohepatitis), fibrosis and cirrhosis are ordinarily determined by resource-intensive and invasive biopsies. We aim to show that routine clinical tests offer sufficient information to predict these endpoints.

Methods: Using the LITMUS Metacohort derived from the European NAFLD Registry, the largest MASLD dataset in Europe, we create three combinations of features which vary in degree of procurement including a 19-variable feature set that are attained through a routine clinical appointment or blood test. This data was used to train predictive models using supervised machine learning (ML) algorithm XGBoost, alongside missing imputation technique MICE and class balancing algorithm SMOTE. Shapley Additive exPlanations (SHAP) were added to determine relative importance for each clinical variable.

Results: Analysing nine biopsy-derived MASLD outcomes of cohort size ranging between 5385 and 6673 subjects, we were able to predict individuals at training set AUCs ranging from 0.719-0.994, including classifying individuals who are At-Risk MASH at an AUC = 0.899. Using two further feature combinations of 26-variables and 35-variables, which included composite scores known to be good indicators for MASLD endpoints and advanced specialist tests, we found predictive performance did not sufficiently improve. We are also able to present local and global explanations for each ML model, offering clinicians interpretability without the expense of worsening predictive performance.

Conclusions: This study developed a series of ML models of accuracy ranging from 71.9-99.4% using only easily extractable and readily available information in predicting MASLD outcomes which are usually determined through highly invasive means.