A computational intelligence tool for the detection of hypertension using empirical mode decomposition

Area of Science:

• Biomedical Engineering
• Computational Intelligence
• Cardiology

Background:

• Hypertension (HPT), or high blood pressure, is a significant risk factor for cardiovascular, cerebrovascular, and renal diseases.
• Masked hypertension (MHPT) presents a diagnostic challenge, as blood pressure appears normal instantaneously but is elevated over 24 hours, leading to delayed or insufficient treatment.

Purpose of the Study:

• To develop and evaluate a computational intelligence tool (CIT) for detecting hypertension (HPT) and masked hypertension (MHPT) using electrocardiogram (ECG) signals.
• To address the diagnostic difficulties associated with MHPT and improve patient management.

Main Methods:

• ECG signals were pre-processed and decomposed into intrinsic mode functions (IMFs) using Empirical Mode Decomposition (EMD) up to five levels.
• Nonlinear features were extracted from the IMFs and a subset of discriminatory features was selected using Student's t-test.
• The selected features were input into various classifiers, with the k-nearest neighbor (k-NN) algorithm achieving the highest accuracy.

Main Results:

• The k-nearest neighbor (k-NN) classifier achieved a diagnostic accuracy of 97.70% for classifying hypertension versus normal ECG signals.
• The developed system demonstrated robust performance, validated through a 10-fold cross-validation technique.

Conclusions:

• The developed computational intelligence tool shows significant promise for the automatic classification of hypertension using ECG signals.
• This system can serve as a valuable diagnostic aid in hospital settings for identifying HPT and potentially MHPT, facilitating timely intervention.