Steatotic LIver Disease: Machine Learning and Multi-Omics for Nutrition Education and Precision Therapy (SLIM-NET)

Steatotic liver disease (SLD), or 'fatty liver disease', is a major health problem in the UK and worldwide. Liver disease is now the second commonest cause of preventable death amongst individuals of working age in the UK; the major risk factors for SLD are alcohol use, obesity, diabetes mellitus, high blood pressure and an altered lipid profile. SLD can lead to advanced liver scarring (fibrosis), and complications such as liver failure or liver cancer. We are increasingly aware that SLD is not a single disease, but a spectrum of conditions with slightly different factors at play. This means that different patients may have different risks of progression to severe liver disease or other health problems, and may respond differently to treatments. Although new drug treatments are being developed, a ‘one-size-fits-all’ approach is unlikely to be effective. Additionally, new drugs are likely to carry a significant cost burden. Therefore, the cornerstone of treatment for SLD is diet, but current dietary advice for SLD is generic and doesn't always work for everyone. The aim of this proposal is to use cutting-edge data analysis techniques (machine learning, ML), and large data sets with information on diet, genetics and profiling of molecules (proteins, metabolites), to transform how we approach dietary management of SLD. We will: Identify distinct SLD subtypes: Using these data from large UK studies (Whitehall II, EPIC-Norfolk, and UK Biobank), we will use ML to identify subtypes of SLD. These subtypes will be distinguished by unique combinations of genetics and exposure to risk factors. Uncover dietary links: We will then analyse detailed dietary information to identify specific dietary patterns associated with each SLD subtype. This will help us understand how different diets modify disease progression in different groups of people. Design tailored dietary interventions: Based on these findings, we will work with nutrition experts and patients to design dietary interventions tailored to the each SLD subtype. These interventions will move beyond general advice and provide specific, actionable recommendations. Develop a Biomarker Panel: We will also develop a new tool to monitor how well people are sticking to their dietary plans. This will involve identifying specific molecules in the blood that are associated with dietary intake. These will need validating testing in future, controlled feeding studies, but potentially allow us to objectively measure dietary compliance. Potential Applications and Benefits: Importantly, this proposal is in line with recommendations from SLD expert panels, and a patient-focussed nationwide survey of liver cirrhosis research priorities led by the James Lind Alliance. This research has the potential to: Improve risk prediction in SLD: Through applying these ML techniques, we will increase understanding of different subtypes of SLD with different health outcomes. This information will improve management of SLD and target use of future treatments and clinical trials. Develop novel dietary interventions for SLD: The proposal will also develop tailored diets for each subtype of SLD. At the end of this project, they will be ready for testing in controlled clinical trials. Advance precision nutrition: This project will contribute to the broader field of precision nutrition by demonstrating how advanced data analysis techniques can be used to tailor dietary advice. We will make all methods and code available for other researchers in the field.