Advanced search
Start date

The use of machine learning to integrate lipidomics, clinical chemistry and functional parameters of normal-weight, obese and type 2 diabetic middle-aged people after 16 weeks of combined training

Grant number: 19/25817-6
Support type:Scholarships abroad - Research Internship - Doctorate
Effective date (Start): October 04, 2020
Effective date (End): August 03, 2021
Field of knowledge:Health Sciences - Physical Education
Principal Investigator:Claudia Regina Cavaglieri
Grantee:Renata Garbellini Duft
Supervisor abroad: Julian Griffin
Home Institution: Faculdade de Educação Física (FEF). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
Local de pesquisa : Imperial College London, England  
Associated to the scholarship:17/11033-8 - Acute and chronic metabolic responses to physical exercise in healthy, obese, and diabetic type 2 middle aged people using metabolomics and lipidomics approaches, BP.DR


Lipidomics is an approach that elucidates the complexity of the mechanisms involved in the maintenance of lipid homeostasis to find candidate biomarkers which in turn may act as diagnostic tools for chronic noncommunicable diseases. The excess amount of adipose tissue in obesity leads to a variety of metabolic dysfunctions, such as insulin resistance (IR) and type 2 diabetes (T2D). If we are to tackle the growing epidemic of obesity, an essential nonpharmacological treatment for this and other comorbidities is the regular practice of physical exercise, including combined training (CT) which combines aerobic training (AT) and resistance training (RT) in the same session. Classical biochemical studies provide a limited insight into the global network of metabolism, and for this reason, new strategies are being used for data integration. For example, machine learning algorithms can be used to interrogate omic datasets. With this in mind, the aim of this project is to use machine learning algorithms to integrate lipidomics data with clinical markers and functional training parameters to find candidate biomarkers that predict metabolic changes caused by training based on the metabolism of normal-weight (NW), obese (OB) and type 2 diabetic (TD2), middle-aged people. The study consists of 127 middle-aged volunteers of both sexes, divided into 3 control groups and 3 training groups, according to their condition. The CT program was performed 3 times a week for 16 weeks. Blood collections and subcutaneous adipose tissue biopsies were collected before the period of training and 48 to 72h after the last training session. Lipidomics analyses will be performed by mass-spectrometry, and the data integration and use of machine learning will be conducted using the statistical tool R and Matlab. (AU)