The application of mechanical ventilation is a life-saving routine therapy that allows the patient to overcome the physiological impact of surgeries, trauma or critical illness by ensuring vital oxygenation and carbon dioxide removal. Above a certain level of minute ventilation (usually set to ensure acceptable carbon dioxide removal and oxygenation) oxygenation is only marginally affected by a further increase in minute ventilation. Thus, oxygenation is predominantly influenced by inspiratory oxygen fraction (FiO2) Usually, finding the appropriate setting is a trial-and-error procedure, as the clinician is unaware of the exact value that needs to be set in order to reach the desired arterial oxygen partial pressures (PaO2) in the patient.

BackgroundSuccessful application of mechanical ventilation as a life-saving therapy implies appropriate ventilator settings. Decision making is based on clinicians’ knowledge, but can be enhanced by mathematical models that determine the individual patient state by calculating parameters that are not directly measurable. Evaluation of models may support the clinician to reach a defined treatment goal. Bedside applicability of mathematical models for decision support requires a robust identification of the model parameters with a minimum of measuring effort. The influence of appropriate data selection on the identification of a two-parameter model of pulmonary gas exchange was analyzed.MethodsThe model considers a shunt as well as ventilation-perfusion-mismatch to simulate a variety of pathologic pulmonary gas exchange states, i.e. different severities of pulmonary impairment. Synthetic patient data were generated by model simulation. To incorporate more realistic effects of measurement errors, the simulated data were corrupted with additive noise. In addition, real patient data retrieved from a patient data management system were used retrospectively to confirm the obtained findings. The model was identified to a wide range of different FiO2 settings. Just one single measurement was used for parameter identification. Subsequently prediction performance was obtained by comparing the identified model predicted oxygen level in arterial blood either to exact data taken from simulations or patients measurements.ResultsStructural identifiability of the model using one single measurement for the identification process could be demonstrated. Minimum prediction error of blood oxygenation depends on blood gas level at the time of system identification i.e. the measurement situation. For severe pulmonary impairment, higher FiO2 settings were required to achieve a better prediction capability compared to less impaired pulmonary states. Plausibility analysis with real patient data could confirm this finding.Discussion and conclusionsDependent on patients’ pulmonary state, the influence of ventilator settings (here FiO2) on model identification of the gas exchange model could be demonstrated. To maximize prediction accuracy i.e. to find the best individualized model with as few data as possible, best ranges of FiO2-settings for parameter identification were obtained. A less effort identification process, which depends on the pulmonary state, can be deduced from the results of this identifiability analysis.

No abstract

Model based decision support helps in optimizing therapy settings for individual patients while providing additional insight into a patient’s disease state through the identified model parameters. Using multiple models with different simulation focus and complexity allows adapting decision support to the current clinical situation and the available data. A previously presented set of numerical criteria allows selecting the best model based on fit quality, model complexity, and how well the parameter values are defined by the presented data. To systematically evaluate those criteria in an algorithm we have created insilico data sets using four different respiratory mechanics models with three different parameter settings each. Each of those artificial patients was ventilated with three different manoeuvres and the resulting data was used to identify the same models used to create the data. The selection algorithm was then presented with the results to select the best model. Not considering determinateness of the identified model parameters, the algorithm chose the same model that was used to create the data in 78%, a more complex model in 5% and a less complex model in 18% of all cases. When including the determinateness of model parameters in the decision process, the algorithm chose the same model in 42% of the cases and a less complex model in 56% of all cases. In 2% of the presented cases, no model complied with the required criteria.

No abstract