Artifact detection and data reconciliation in multivariate ventilatory variables measured during anesthesia: A case study

A ventilation system is usually connected to an anesthetized patient during surgery to replace or support normal respiratory function. Clinician experts monitor the concentration, flow and pressure of the various gases in the airway to maintain adequate ventilation. However, environmental disturbances often perturb the readings of these variables and cause them to deviate far from the true levels, resulting in a biased evaluation of a patient´s ventilatory status. Most of the previously proposed signal estimation methods, however, have only utilised the difference of the dynamic characteristics between artifacts and the true physiological variations in each individual variable, without considering the interrelationship between these signals. This has resulted in suboptimal signal estimation. In this paper, we propose the use of the procedure of gross error detection and data reconciliation commonly used in process control, to detect highly cross-correlated artifacts in the ventilation circuit and reconcile the uncontaminated signal measurements. A case study demonstrates that the proposed method has great potential for improving the reliability of ventilatory signals.