Breath Analyzer\nfor Real-Time Exercise Fat Burning\nPrediction: Oral and Alveolar Breath Insights with CNN

The increasing prevalence of obesity and metabolic disorders\nhas\ncreated a significant demand for personalized devices that can effectively\nmonitor fat metabolism. In this study, we developed an advanced breath\nanalyzer system designed to provide real-time monitoring of exercise-induced\nfat burning by analyzing volatile organic compounds (VOCs) present\nin both oral and alveolar breath. Acetone in exhaled breath and β-hydroxybutyric\nacid (BOHB) in the blood are both biomarkers closely linked to the\nmetabolic fat burning process occurring in the liver, particularly\nafter exercise. The breath analyzer utilizes a sensor array to detect\nVOC patterns, with the data analyzed using a one-dimensional convolutional\nneural network (1D CNN) for an accurate prediction of BOHB levels\nin the blood. We collected and analyzed 30 exhaled breath samples\nwith our analyzer and blood samples for BOHB from participants before\nand after exercise. The results showed a strong correlation between\nsensor responses and BOHB levels, with Pearson correlation coefficients\nof 0.99 across different postexercise time points. The 1D CNN model\neffectively estimated BOHB concentrations, achieving Pearson coefficients\nof 0.96 for the training data set and 0.86 for the test data set.\nAdditionally, our findings confirm that alveolar air samples, which\ncontain metabolic byproducts from deeper in the lungs, offer more\nreliable data for fat burning analysis than oral air samples. This\nnoninvasive, real-time breath monitoring tool offers a promising solution\nfor individuals demanding to optimize their exercise routines and\ntrack metabolic health with high precision and accuracy.