Feature Selection for Hypertension Risk Prediction Using XGBoost on Single Nucleotide Polymorphism Data.

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Hypertension is a global health concern with severe complications if untreated.
• Identifying reliable biomarkers for hypertension risk is crucial for early detection and prevention.
• Genetic variations, specifically single nucleotide polymorphisms (SNPs), offer potential markers for disease predisposition.

Purpose of the Study:

• To develop a feature selection model using the XGBoost algorithm.
• To identify specific single nucleotide polymorphisms (SNPs) as effective biomarkers for hypertension risk detection.
• To evaluate the performance of the XGBoost model in predicting hypertension.

Main Methods:

• Utilized the OpenSNP dataset comprising 19,697 SNPs from 2,052 samples.
• Employed Extreme Gradient Boosting (XGBoost), an ensemble machine learning method, for feature selection.
• Built a classifier model using high-dimensional genetic variation data (SNPs) for prediction.

Main Results:

• Identified 292 significant SNPs for hypertension risk prediction.
• Achieved high performance metrics: 98.55% F1-score, 98.73% precision, 98.38% recall, and 98% overall accuracy.
• Demonstrated superior performance of XGBoost feature selection compared to genetic algorithms, ANOVA, chi-square, and PCA.

Conclusions:

• Successfully developed a predictive model for hypertension using SNP data.
• Effectively managed high-dimensional SNP data to pinpoint significant features as biomarkers.
• The XGBoost feature selection method shows high efficacy in predicting hypertension risk.