Identifying Coronary Artery Lesions by Feature Analysis of Radial Pulse Wave: A Case-Control Study

Area of Science:

• Cardiology
• Biomedical Engineering
• Data Science

Background:

• Cardiovascular diseases (CVDs) are a leading global cause of mortality.
• Invasive coronary angiography, while accurate, is not optimal for all suspected coronary heart disease (CHD) patients.
• Radial pulse wave analysis presents a promising noninvasive technique for cardiovascular assessment.

Purpose of the Study:

• To analyze radial pulse waves for models assessing coronary artery lesion extent.
• To investigate the potential of noninvasive pulse wave technology for CVD evaluation.
• To support the development of wearable devices and mobile medicine for cardiovascular health.

Main Methods:

• 529 patients with suspected CHD underwent coronary angiography and radial pulse wave analysis.
• Radial pulse wave signals were analyzed for linear time-domain and nonlinear multiscale entropy features.
• Machine learning algorithms (KNN, DT, RF) were employed to build models for lesion identification.

Main Results:

• Specific pulse wave features differed significantly between control, 1-2 lesion, and multiple lesion groups (P < 0.05).
• The Random Forest (RF) model demonstrated higher average precision compared to KNN and DT.
• A combined model using time-domain and multiscale entropy features achieved the highest average precision of 80.98%.

Conclusions:

• Radial pulse wave analysis can accurately identify the range of coronary artery lesions.
• This noninvasive technique holds significant value for assessing coronary artery lesion severity.
• The findings support the development of mobile medical treatments and remote monitoring for at-risk patients.