Validity and reliability of seismocardiography for the estimation of cardiorespiratory fitness

Background: Low cardiorespiratory fitness (ie, peak oxygen consumption [V.O₂peak]) is associated with cardiovascular disease and all-cause mortality and is recognized as an important clinical tool in the assessment of patients. Cardiopulmonary exercise test (CPET) is the gold standard procedure for determination of V.O₂peak but has methodological challenges as it is time-consuming and requires specialized equipment and trained professionals. Seismofit is a chest-mounted medical device for estimating V.O₂peak at rest using seismocardiography. Objective: The purpose of this study was to investigate the validity and reliability of Seismofit V.O₂peak estimation in a healthy population. Methods: On 3 separate days, 20 participants (10 women) underwent estimations of V.O₂peak with Seismofit (×2) and Polar Fitness Test (PFT) in randomized order and performed a graded CPET on a cycle ergometer with continuous pulmonary gas exchange measurements. Results: Seismofit V.O₂peak showed a significant bias of –3.1 ± 2.4 mL·min^–1·kg^–1 (mean ± 95% confidence interval) and 95% limits of agreement (LoA) of ±10.8 mL·min^–1·kg^–1 compared to CPET. The mean absolute percentage error (MAPE) was 12.0%. Seismofit V.O₂peak had a coefficient of variation of 4.5% ± 1.3% and an intraclass correlation coefficient of 0.95 between test days and a bias of 0.0 ± 0.4 mL·min^–1·kg^–1 with 95% LoA of ±1.6 mL·min^–1·kg^–1 in test–retest. In addition, Seismofit showed a 2.4 mL·min^–1·kg^–1 smaller difference in 95% LoA than PFT compared to CPET. Conclusion: The Seismofit is highly reliable in its estimation of V.O₂peak. However, based on the measurement error and MAPE >10%, the Seismofit V.O₂peak estimation model needs further improvement to be considered for use in clinical settings.