Estimation of Respiratory Nasal Pressure and Flow Rate Signals Using Different Respiratory Sound Features - 07/12/22
Abstract |
Background |
Respiratory sounds are associated with the flow rate, nasal flow pressure, and physical characteristics of airways. In this study, we aimed to develop the flow rate and nasal flow pressure estimation models for the clinical application, and find out the optimal feature set for estimation to achieve the optimal model performance.
Methods |
Respiratory sounds and flow rate were acquired from nine healthy volunteers. Respiratory sounds and nasal flow pressure were acquired from twenty-three healthy volunteers. Four types of respiratory sound features were extracted for flow rate and nasal flow pressure estimation using different estimation models. Effects of estimations using these features were evaluated using Bland-Altman analysis, estimation error, and respiratory sound feature calculation time. Besides, expiratory and inspiratory phases divided estimation errors were compared with united estimation errors.
Results |
The personalized logarithm model was selected as the optimal flow rate estimation model. Respiratory nasal flow pressure estimation based on this model was also performed. For the four different respiratory sound features, there is no statistically significant difference in flow rate and pressure estimation errors. LogEnvelope was, therefore, chosen as the optimal feature because of the lowest computational cost. In addition, for any type of respiratory sound feature, no statistically significant difference was observed between divided and united estimation errors (flow rate and pressure).
Conclusion |
Respiratory flow rate and nasal flow pressure can be estimated accurately using respiratory sound features. Expiratory and inspiratory phases united estimation using respiratory sounds is a more reasonable estimation method than divided estimation. LogEnvelope can be used for this united respiratory flow rate and nasal flow pressure estimation with minimum computational cost and acceptable estimation error.
Le texte complet de cet article est disponible en PDF.Graphical abstract |
Highlights |
• | Respiratory flow rate and nasal flow pressure can be estimated accurately using respiratory sound features. |
• | Personalization model is the optimal model for estimation. |
• | Log of the envelope is select as the optimal feature for minimum computational cost and acceptable estimation error. |
• | No statistically difference between errors of divided and united estimation. |
Keywords : Respiratory sound features, Respiratory flow rate estimation, Respiratory nasal pressure estimation, Expiratory and inspiratory phases divided and united estimation
Plan
Vol 43 - N° 6
P. 694-704 - décembre 2022 Retour au numéroBienvenue sur EM-consulte, la référence des professionnels de santé.
L’accès au texte intégral de cet article nécessite un abonnement.
Déjà abonné à cette revue ?