Integrated Multi-omics and Machine Learning-driven Personalized Treatment of Obesity-associated Fatty Liver Disease

The investigators study the most prevalent liver disease in the history of humankind, which is the leading cause of liver transplantation in its severe forms. It results from two silent pandemics with enormous health impacts: obesity and diabetes. Together or separately, they affect more than 30% of the world's population. The current term for the disease is MAFLD (metabolic (dysfunction)-associated fatty liver disease). This designation indicates that metabolic disorders related to obesity, diabetes, dyslipidemia, and hypertension are its primary cause. These disorders are related and lead to fat accumulation in the liver, the first step in a broad spectrum of chronic liver diseases. These diseases respond clinically in a very variable way and remain undiagnosed and untreated for a long time. There is no accepted pharmacological treatment, and lifestyle changes, although possibly effective, usually fail because they require particularly favorable conditions. Therefore, the identified problems that should be solve are:

(1) The diagnosis of MAFLD requires a liver biopsy, a costly and aggressive procedure. (2) Without examining the liver, clinicians can know little about the progression of the disease and the underlying causes. (3) The results in experimental models can be informative but difficult to translate to the clinic. Recent reports suggest the essential role of phospholipid biosynthesis and transport between the endoplasmic reticulum and mitochondria. (4) All of the above makes it difficult to obtain the necessary information to propose changes in clinical guidelines.

Considering these aspects, patients with morbid obesity can be an informative human model. Among other advantages, patients have surgical options that allow us to obtain portions of affected organs that facilitate specific diagnosis and that, because they require constant care, can be studied on an ongoing basis. The presented approach can improve patient care and essentially consists of identifying the most significant number of variables that can help. In particular, here are proposed the inclusion of variables that can already be obtained from recent advances in the laboratory, encompassed within the omics sciences (genomics, transcriptomics, proteomics, metabolomics, lipidomics, microbiomics). Each of these has its advantages and limitations. Predictive models can integrate these variables into clinical data to explore organ crosstalk.