Development of deep-learning models for real-time anaerobic threshold and peak VO2 prediction during cardiopulmonary exercise testing

• Published in: European journal of preventive cardiology Vol. 31; no. 4; pp. 448 - 457
• Main Authors: Watanabe, Tatsuya, Tohyama, Takeshi, Ikeda, Masataka, Fujino, Takeo, Hashimoto, Toru, Matsushima, Shouji, Kishimoto, Junji, Todaka, Koji, Kinugawa, Shintaro, Tsutsui, Hiroyuki, Ide, Tomomi
• Format: Journal Article
• Language: English
• Published: US Oxford University Press 04.03.2024
• Subjects: Deep learning; Anaerobic threshold; Peak oxygen uptake; Cardiopulmonary exercise testing; Respiratory gas analysis; deep-learning; peak oxygen uptake; cardiopulmonary exercise testing; respiratory gas analysis; anaerobic threshold
• ISSN: 2047-4873; 2047-4881
• DOI: 10.1093/eurjpc/zwad375

Abstract Aims Exercise intolerance is a clinical feature of patients with heart failure (HF). Cardiopulmonary exercise testing (CPET) is the first-line examination for assessing exercise capacity in patients with HF. However, the need for extensive experience in assessing anaerobic threshold (AT) and the potential risk associated with the excessive exercise load when measuring peak oxygen uptake (peak VO2) limit the utility of CPET. This study aimed to use deep-learning approaches to identify AT in real time during testing (defined as real-time AT) and to predict peak VO2 at real-time AT. Methods and results This study included the time-series data of CPET recorded at the Department of Cardiovascular Medicine, Kyushu University Hospital. Two deep neural network models were developed to: (i) estimate the AT probability using breath-by-breath data and (ii) predict peak VO2 using the data at the real-time AT. The eligible CPET contained 1472 records of 1053 participants aged 18–90 years and 20% were used for model evaluation. The developed model identified real-time AT with 0.82 for correlation coefficient (Corr) and 1.20 mL/kg/min for mean absolute error (MAE), and the corresponding AT time with 0.86 for Corr and 0.66 min for MAE. The peak VO2 prediction model achieved 0.87 for Corr and 2.25 mL/kg/min for MAE. Conclusion Deep-learning models for real-time CPET analysis can accurately identify AT and predict peak VO2. The developed models can be a competent assistant system to assess a patient’s condition in real time, expanding CPET utility. Lay Summary Cardiopulmonary exercise testing can be used to evaluate the condition of patients with heart failure during exercise. Developed deep-learning models can accurately predict a patient’s anaerobic threshold in real time and peak oxygen uptake. The models can be used by clinicians for more objective and accurate assessments in real time, expanding the utility of cardiopulmonary exercise testing. Graphical Abstract Graphical Abstract