Characterizing the critical features when personalizing antihypertensive drugs using spectrum analysis and machine...

Area of Science:

• Cardiology
• Pharmacology
• Medical Informatics

Background:

• Current hypertension treatment faces challenges in prescribing optimal drugs based on individual patient characteristics.
• Identifying specific clinical features associated with drug effectiveness is crucial for efficient hypertension management.

Purpose of the Study:

• To apply data mining techniques to determine the clinical characteristic spectrum for five common antihypertensive drugs.
• To extract critical clinical features influencing the efficacy of Irbesartan, Metoprolol, Felodipine, Amlodipine, and Levamlodipine.

Main Methods:

• Utilized spectrum analysis based on statistical methods and five machine learning algorithms.
• Compared successful and unsuccessful treatment cases to identify significant clinical features.
• Developed a visualized relative weight matrix by integrating statistical and machine learning results.

Main Results:

• Identified distinct importance orders for 15 clinical features across the five antihypertensive agents.
• The extracted clinical attributes for each drug were found to be clinically reasonable and meaningful.
• Demonstrated that different hypertension drugs are associated with varying sets of critical patient features.

Conclusions:

• The study provides a data-driven reference for personalized antihypertensive drug selection.
• The identified clinical features can aid physicians in matching patients with the most effective hypertension medications.
• This approach enhances the precision of hypertension treatment by considering individual patient profiles.