Comparison of dimension reduction methods on fatty acids food source study

Area of Science:

• Nutritional Biochemistry
• Statistical Analysis in Biology
• Epidemiology

Background:

• Serum fatty acids (FAs) are distributed across triglyceride (TG), phospholipid (PL), cholesteryl ester (CE), and free fatty acid (FFA) fractions.
• Understanding the dietary sources of serum FAs is crucial for nutritional and epidemiological research.

Purpose of the Study:

• To evaluate and compare four statistical methods (ICA, factor analysis, CPCA, PCA) for identifying serum FA food sources.
• To determine the efficacy of Independent Component Analysis (ICA) in distinguishing dietary fat sources, including animal fat, from endogenous FAs.

Main Methods:

• Application of Independent Component Analysis (ICA), Factor Analysis, Common Principal Component Analysis (CPCA), and Principal Component Analysis (PCA) to serum lipid fractions.
• Specific analysis of FFA variables using ICA and Factor Analysis to differentiate animal fat intake from endogenous FAs.
• Correlation analysis between published food FA composition data and ICA loading values.

Main Results:

• Independent Component Analysis (ICA) yielded the most informative results for identifying serum FA food sources.
• ICA successfully distinguished between endogenous FAs, plant oil intake, animal fat intake, and fish oil intake.
• ICA demonstrated a novel capability to differentiate animal FAs from endogenous FAs, outperforming Factor Analysis.

Conclusions:

• Independent Component Analysis (ICA) is a valuable and effective tool for uncovering the dietary sources of serum fatty acids.
• The findings suggest ICA has significant potential for improving epidemiological studies by accurately identifying FA origins.
• Specific food sources contributing to serum FA profiles can be identified using ICA in conjunction with food composition databases.