Predicting High Blood Pressure Using DNA Methylome-Based Machine Learning Models.

Area of Science:

• Epigenetics
• Computational Biology
• Cardiovascular Disease Research

Background:

• DNA methylation is crucial in high blood pressure (BP) pathophysiology.
• Early detection of high BP is vital for effective management and prevention of complications.

Purpose of the Study:

• To investigate the utility of DNA methylome data for detecting high BP using machine learning (ML).
• To compare the performance of deep learning (DL), support vector machine, and random forest algorithms in BP prediction.
• To identify specific methylation sites associated with BP measures.

Main Methods:

• Applied DL, support vector machine, and random forest algorithms to DNA methylome data from 50 elderly individuals.
• Utilized a nested five-group cross-validation approach for performance evaluation.
• Employed data normalization, dimensionality reduction (principal component analysis), and mixed-effects models.

Main Results:

• Deep learning (DL) demonstrated the best predictive performance with an AUROC of 0.73 and accuracy of 0.69.
• Identified 61,694 intragenic and 16,754 intergenic methylation sites significantly associated with BP measures.
• The study pioneered the application of ML and DNA methylome for early high BP detection.

Conclusions:

• ML algorithms, particularly DL, can effectively detect high BP using DNA methylome profiles.
• DNA methylome data holds potential as a biomarker for early high BP detection.
• This methodology offers a novel approach for clinical high BP screening and management.