Machine Learning-Assisted Janus Colorimetric Face Mask for Breath Ammonia Analysis

Area of Science:

• Biomedical Engineering
• Chemical Sensing
• Wearable Technology

Background:

• Artificial olfactory systems are crucial for analyzing volatile organic compounds (VOCs) in exhaled breath for medical applications.
• A need exists for portable, accurate breath VOC analysis systems in the healthcare industry.
• Current methods often lack the portability and comfort required for widespread clinical use.

Purpose of the Study:

• To develop a wearable and portable sensing system for comfortable evaluation of breath ammonia levels.
• To integrate colorimetry, Janus fabric design, and machine learning for breath analysis.
• To create a system for real-time monitoring of ammonia concentrations in exhaled breath.

Main Methods:

• A Janus colorimetric face mask (JCFM) was designed for unidirectional moisture penetration, ensuring facial comfort.
• Four pH indicators on a colorimetric array were used as recognition elements to detect ammonia.
• The K-nearest neighbor (K-NN) machine learning algorithm was employed for ammonia level analysis based on RGB response features.

Main Results:

• The JCFM demonstrated a linear relationship between ammonia concentration (1-10 ppm) and Euclidean distance (ED) response (R^2 = 0.988).
• The K-NN algorithm achieved a prediction accuracy of 96% for breath ammonia levels.
• The system effectively captured optical fingerprint information of breath ammonia, mimicking mammalian olfactory structures.

Conclusions:

• The developed JCFM provides a comfortable, wearable, and portable solution for breath ammonia analysis.
• Integration of colorimetry, Janus design, and machine learning enables accurate and efficient VOC detection.
• This sensing system holds significant potential for non-invasive medical diagnostics and healthcare monitoring.