Work

Identification of Novel Biomarkers by Multiomics Data Integration

Year: 2023
Type: dissertation
Abstract: Non-alcoholic fatty liver disease (NAFLD) is a widespread liver disease caused by an accumulation of excess liver fat which can lead to various severe complications and permanent liver damage. NAFLD develops in several steps, beginning with hepatic steatosis (HS). Some underlying risk factors for NAFLD are genetics, gut microbiome, diet, and metabolic diseases. Early detection of HS can dramatically increase positive outlook; however, existing diagnostic methods are often invasive, expensive, or otherwise inaccessible. This project aimed to use machine learning to create a model that can be used for more accessible diagnosis of HS. Based on a previous study by Zeybel et al., a Python program was written to analyze the multiomics data set from the study. The multiomics data was obtained from 56 obese patients diagnosed with HS of varying severity. Follow-up data was gathered from 22 of these patients after three months. The chosen machine learning algorithm was random forest and the program analyzed the data from the previous study with the goal of predicting whether a patient had HS and, in that case, at what stage. This was done by calculating the p values for each biomarker and then choosing combinations of some of them to use as models for predicting the disease. We found that the highest accuracy of diagnosis was provided by a model based on five lipid biomarkers as well as five clinical biomarkers at a diagnostic accuracy of 82.4%. These results are promising for future diagnostic tools; however, more research is needed for this kind of diagnostic tool to be seen as reliable.
Authors: Wijk, Ellinor, Sandell, Erika, Sjöblom, Gabriel, Eriksson, Jonas, Kahilainen, Melissa
Language: English
Subfield: Hepatology
Field: Medicine
Open Access status: green
OpenAlex ID: W7038369130
Cites: 22
Cited by:
Related to: 10