Exhaled breath performance in diagnosing ischemic heart disease utilizing a machine learning model.
Basheer Abdullah Marzoog1, Anastasia Stroeva1, Malika Mustafina2
1Institute of Personalized Cardiology of The Center 'Digital Biodesign and Personalized Healthcare' of Biomedical Science and Technology Park, Sechenov First Moscow State Medical University, 8-2 Trubetskaya street, 119991 Moscow, Moscow, Russia.
Journal of Breath Research
|May 18, 2026
Summary
Machine learning models analyzing exhaled breath volatile organic compounds (VOCs) show promise for detecting ischemic heart disease (IHD). While not a standalone tool, this non-invasive breath analysis may aid in diagnosing IHD.
Area of Science:
- Cardiology
- Computational Biology
- Analytical Chemistry
Background:
- Ischemic heart disease (IHD) is a leading cause of death globally.
- Current diagnostic methods for IHD are limited by cost, accessibility, invasiveness, and accuracy.
- There is a critical need for novel, non-invasive screening techniques for IHD.
Purpose of the Study:
- To assess the efficacy of a machine learning model utilizing dynamic exhaled breath volatile organic compound (VOC) patterns.
- To detect stress-induced myocardial perfusion defects, a key indicator of ischemic heart disease.
- To evaluate the diagnostic potential of breath VOC analysis as a non-invasive screening tool for IHD.
Main Methods:
- Prospective single-center study involving 80 participants (31 with confirmed myocardial perfusion defects, 49 controls).
- Real-time breath analysis using PTR-TOF-MS-1000 at rest and post-exercise.
- Machine learning models developed using changes in VOC patterns, validated with leave-one-out cross-validation.
Main Results:
- The machine learning model achieved an AUC of 0.743, with a sensitivity of 0.774 and specificity of 0.633.
- Key VOCs (m/z 94.053731, 90.951122, 60.055305) were identified as significant diagnostic features.
- Dynamic breath analysis demonstrated feasibility in detecting stress-induced myocardial perfusion defects.
Conclusions:
- Dynamic exhaled breath analysis combined with machine learning shows potential for ischemic heart disease detection.
- The current model, validated internally, is not suitable as a standalone screening tool due to sensitivity limitations.
- Further external validation is necessary for clinical implementation as an adjunctive or triage-support tool.
