A novel method for dipper/non-dipper pattern classification in hypertensive and non-diabetic patients

Area of Science:

• Cardiology
• Medical Informatics

Background:

• Ambulatory blood pressure monitoring (ABPM) data are crucial for dipper/non-dipper pattern classification.
• Physicians and algorithms rely on ABPM for accurate diagnosis.
• A need exists to simplify the classification process by potentially excluding ABPM data.

Purpose of the Study:

• To develop a classification model for dipper/non-dipper patterns in hypertensive, non-diabetic patients.
• To achieve high performance metrics while excluding ABPM data.
• To identify alternative, simpler diagnostic attributes.

Main Methods:

• Utilized data from 29 hypertensive patients without diabetes.
• Applied artificial neural network algorithms for classification.
• Focused on attribute reduction, excluding ABPM data.

Main Results:

• The artificial neural network model successfully classified dipper/non-dipper patterns.
• Achieved a highest accuracy of 87.5%, sensitivity of 71%, and specificity of 94%.
• Identified Ewing-score and HRREP as key predictive attributes.

Conclusions:

• A novel, simplified method for dipper/non-dipper classification was developed.
• This method uses only Ewing-score and HRREP, offering a fast and low-cost alternative.
• The attribute reduction approach shows potential for broader applications in disease diagnosis with large datasets.