Objectives: The goal of the current study is to examine the diagnostic usefulness of lysosomal acid lipase (LAL) activity in cirrhotic patients according to healthy control groups utilizing assessment methods including machine learning methods.
Methods: An open-access dataset was used in the current study and included 63 patients with cryptogenic cirrhosis (cryptogenic group), 88 patients with cirrhosis of known underlying causes (non-cryptogenic group), and 97 healthy individuals (control group). Mann Whitney U test was utilized for comparisons between cirrhosis groups and control group and for comparisons between cirrhosis groups. The Stochastic Gradient Boosting (SGB) method was utilized to model the patients with cirrhosis and the control group. Results: LAL activity was observed to be reduced in cirrhotic patients (cryptogenic and known etiology cirrhosis) when compared to control groups. The machine learning modelling algorithm showed that the cirrhotic group and the control group were distinguished from each other with an accuracy of 93.2%. The platelets, LAL activity, ALT, AST, and white blood cells were obtained as the most important variables according to variable importance values.
Conclusion: LAL activity value can be evaluated diagnostic marker in cirrhotic patients. However, it is not a discriminative marker that differentiates different etiologies (cryptogenic and cirrhosis of known etiology).