A Novel Pulse-Taking Device for Persian Medicine Based on Convolutional Neural Networks

Area of Science:

• Integrative Medicine
• Biomedical Engineering
• Artificial Intelligence in Healthcare

Background:

• Persian medicine (PM) relies on wrist pulse measurement for health and temperament assessment.
• Subjectivity in physician interpretation of pulse wave features limits integration with modern medicine.
• This paper presents a system to objectively measure pulse signals based on PM.

Purpose of the Study:

• To develop and clinically implement a system for measuring wrist pulse signals aligned with Persian medicine principles.
• To address the challenge of subjective interpretation in traditional pulse diagnosis.
• To enhance the reliability and repeatability of PM-based diagnostic outcomes.

Main Methods:

• A customized device was designed and utilized to log wrist pulse wave data.
• Seven convolutional neural networks (CNNs) were employed for pulse wave feature classification.
• Pulse wave features (pace, rate, elasticity, strength, width, length, height) were assessed by a specialist on 34 participants.

Main Results:

• The system successfully measured pulse wave features according to PM principles.
• Each participant's pulse characteristics were assessed using seven CNNs.
• Participants were classified into three distinct categories based on their pulse assessments.

Conclusions:

• A customized device combined with deep learning algorithms can accurately measure pulse wave features based on PM.
• This technological integration significantly improves the reliability and repeatability of PM diagnostic results.
• The system offers a pathway for integrating traditional Persian medicine with modern medical diagnostics.