BROWSE
Detail
Journal Article
Breath Analyzer for Real-Time Exercise Fat Burning Prediction: Oral and Alveolar Breath Insights with CNN
Cited 0 time in 
Share 
Share
- Authors
- Byeongju Lee, Junyeong Lee, Hyung-Kun Lee, HyungJu Park, Myung-Joon Kwack, Do Yeob Kim, Inkyu Park, Soo Lim, Dae-Sik Lee
- Issue Date
- 2025-04
- Citation
- ACS SENSORS, v.10, no.4, pp.2510-2519
- ISSN
- 2379-3694
- Publisher
- AMER CHEMICAL SOC
- Language
- English
- Type
- Journal Article
- Abstract
- The increasing prevalence of obesity and metabolic disorders has created a significant demand for personalized devices that can effectively monitor fat metabolism. In this study, we developed an advanced breath analyzer system designed to provide real-time monitoring of exercise-induced fat burning by analyzing volatile organic compounds (VOCs) present in both oral and alveolar breath. Acetone in exhaled breath and β-hydroxybutyric acid (BOHB) in the blood are both biomarkers closely linked to the metabolic fat burning process occurring in the liver, particularly after exercise. The breath analyzer utilizes a sensor array to detect VOC patterns, with the data analyzed using a one-dimensional convolutional neural network (1D CNN) for an accurate prediction of BOHB levels in the blood. We collected and analyzed 30 exhaled breath samples with our analyzer and blood samples for BOHB from participants before and after exercise. The results showed a strong correlation between sensor responses and BOHB levels, with Pearson correlation coefficients of 0.99 across different postexercise time points. The 1D CNN model effectively estimated BOHB concentrations, achieving Pearson coefficients of 0.96 for the training data set and 0.86 for the test data set. Additionally, our findings confirm that alveolar air samples, which contain metabolic byproducts from deeper in the lungs, offer more reliable data for fat burning analysis than oral air samples. This noninvasive, real-time breath monitoring tool offers a promising solution for individuals demanding to optimize their exercise routines and track metabolic health with high precision and accuracy.
- KSP Keywords
- Accurate prediction, Blood sample, Breath monitoring, CNN model, Convolution neural network(CNN), Correlation Coefficient, Data sets, Fat metabolism, Metabolic disorders, Metabolic health, Monitoring tool