7M-Multiomic Multiscale and Multifidelity Modeling via Machine Learning: Application to Diagnosis and to Macrovascular and Microvascular complications in NAFLD.
Final Report Abstract
Non-alcoholic fatty liver disease (NAFLD) affects 25–30% of the general population and is characterized by the presence of non-alcoholic fatty liver (NAFL) that can progress to nonalcoholic steatohepatitis (NASH), liver fibrosis, cirrhosis and hepatocellular carcinoma. People with NASH and liver fibrosis demonstrate increased mortality due to higher cardiovascular- and liver-related deaths. To date, liver biopsy is the gold standard for the diagnosis of NASH and for staging liver fibrosis. However, this is an expensive method that carries certain risks for the patient. With the use of sophisticated high throughput procedures, machine learning techniques and by performing a series of human physiology studies, we investigated a large number of molecules (lipids, glycans, hormones) and identified the ones that their blood levels correlate strongly with the severity of the disease. Subsequently, based on these molecules, we managed to develop novel and highly accurate models for diagnosing simultaneously and with a single blood draw the presence of healthy liver status, NAFL or NASH and for assessing the presence (or not) of liver fibrosis. The sensitivity and specificity of these diagnostic models are above 90%. Future research should aim to validate these models in large independent cohorts with heterogenous populations as well as simplify them and improve their predictive accuracy by including additional clinical, genetic or biochemical parameters related to the disease. Altogether, we developed non-invasive models based on a single blood draw that can serve as a low-risk, cost-effective and highly accurate alternative method to liver biopsy for diagnosing and staging NAFLD.
Publications
- Omics, big data and machine learning as tools to propel understanding of biological mechanisms and to discover novel diagnostics and therapeutics. Metabolism. 2018;87:A1-A9
Perakakis N, Yazdani A, Karniadakis GE, Mantzoros C
(See online at https://doi.org/10.1016/j.metabol.2018.08.002) - Follistatins in glucose regulation in healthy and obese individuals. Diabetes Obes Metab. 2019;21(3):683-90
Perakakis N, Kokkinos A, Peradze N, Tentolouris N, Ghaly W, Tsilingiris D, Alexandrou A, Mantzoros CS
(See online at https://doi.org/10.1111/dom.13572) - Non-invasive diagnosis of non-alcoholic steatohepatitis and fibrosis with the use of omics and supervised learning: A proof of concept study. Metabolism. 2019;101:154005
Perakakis N, Polyzos SA, Yazdani A, Sala-Vila A, Kountouras J, Anastasilakis AD, Mantzoros CS
(See online at https://doi.org/10.1016/j.metabol.2019.154005) - Targeted Analysis of Three Hormonal Systems Identifies Molecules Associated with the Presence and Severity of NAFLD. J Clin Endocrinol Metab. 2020;105(3)
Polyzos SA, Perakakis N, Boutari C, Kountouras J, Ghaly W, Anastasilakis AD, Karagiannis A, Mantzoros CS
(See online at https://doi.org/10.1210/clinem/dgz172)
