Potential harmful effects of ventilation therapy could be reduced by model-based predictions of the effects of ventilator settings to the patient. To obtain optimal predictions, the model has to be individualized based on patients' data. Given a nonlinear model, the result of parameter identification using iterative numerical methods depends on initial estimates. In this work, a feasible hierarchical identification process is proposed and compared to the commonly implemented direct approach with randomized initial values. The hierarchical approach is exemplarily illustrated by identifying the viscoelastic model (VEM) of respiratory mechanics, whose a priori identifiability was proven. To demonstrate its advantages over the direct approach, two different data sources were employed. First, correctness of the approach was shown with simulation data providing controllable conditions. Second, the clinical potential was evaluated under realistic conditions using clinical data from 13 acute respiratory distress syndrome (ARDS) patients. Simulation data revealed that the success rate of the direct approach exponentially decreases with increasing deviation of the initial estimates while the hierarchical approach always obtained the correct solution. The average computing time using clinical data for the direct approach equals 4.77 s (SD  =  1.32) and 2.41 s (SD  =  0.01) for the hierarchical approach. These investigations demonstrate that a hierarchical approach may be beneficial with respect to robustness and efficiency using simulated and clinical data.

Mathematical models of respiratory mechanics can offer substantial insight into patient state and pulmonary dynamics that are not directly measurable. Thus, they offer significant potential to evaluate and guide patient-specific lung protective ventilator strategies for Acute Respiratory Distress Syndrome (ARDS) patients. To assure bedside-applicability, the model has to be computationally efficient and identifiable from the available data, while also capturing dominant dynamics observed in ARDS patients.In this work, a recruitment model is enhanced by considering alveolar distension and implemented in a time-continuous respiratory mechanics model. A hierarchical gradient decent approach is used to fit the model to low-flow test responses of 12 ARDS patients.The reported parameter values were physiologically plausible and capable of reproducing the measured pressure responses with high accuracy. Structural identifiability of the model is proven, but a practical identifiability analysis of the results shows a lack of convexity on the error-surface. Covariance analyses reveal limited influence of particular model parameters during parameter identification indicating that successful parameter identification is currently not assured in all test sets.Overall, the presented model is physiologically and clinically relevant, captures ARDS dynamics, and uses clinically descriptive parameters. The patient-specific models show its ability to capture pulmonary dynamics directly relevant to patient condition and clinical guidance. These characteristics can currently not be directly measured or established without such a validated model.

Mechanical ventilation carries the risk of ventilator-induced-lung-injury (VILI). To minimize the risk of VILI, ventilator settings should be adapted to the individual patient properties. Mathematical models of respiratory mechanics are able to capture the individual physiological condition and can be used to derive personalized ventilator settings. This paper presents model-based calculations of inspiration pressure (pI), inspiration and expiration time (tI, tE) in pressure-controlled ventilation (PCV) and a retrospective evaluation of its results in a group of mechanically ventilated patients. Incorporating the identified first order model of respiratory mechanics in the basic equation of alveolar ventilation yielded a nonlinear relation between ventilation parameters during PCV. Given this patient-specific relation, optimized settings in terms of minimal pI and adequate tE can be obtained. We then retrospectively analyzed data from 16 ICU patients with mixed pathologies, whose ventilation had been previously optimized by ICU physicians with the goal of minimization of inspiration pressure, and compared the algorithm's 'optimized' settings to the settings that had been chosen by the physicians. The presented algorithm visualizes the patient-specific relations between inspiration pressure and inspiration time. The algorithm's calculated results highly correlate to the physician's ventilation settings with r = 0.975 for the inspiration pressure, and r = 0.902 for the inspiration time. The nonlinear patient-specific relations of ventilation parameters become transparent and support the determination of individualized ventilator settings according to therapeutic goals. Thus, the algorithm is feasible for a variety of ventilated ICU patients and has the potential of improving lung-protective ventilation by minimizing inspiratory pressures and by helping to avoid the build-up of clinically significant intrinsic positive end-expiratory pressure.

BackgroundPatient-specific respiratory mechanics models can support the evaluation of optimal lung protective ventilator settings during ventilation therapy. Clinical application requires that the individual’s model parameter values must be identified with information available at the bedside. Multiple linear regression or gradient-based parameter identification methods are highly sensitive to noise and initial parameter estimates. Thus, they are difficult to apply at the bedside to support therapeutic decisions.MethodsAn iterative integral parameter identification method is applied to a second order respiratory mechanics model. The method is compared to the commonly used regression methods and error-mapping approaches using simulated and clinical data. The clinical potential of the method was evaluated on data from 13 Acute Respiratory Distress Syndrome (ARDS) patients.ResultsThe iterative integral method converged to error minima 350 times faster than the Simplex Search Method using simulation data sets and 50 times faster using clinical data sets. Established regression methods reported erroneous results due to sensitivity to noise. In contrast, the iterative integral method was effective independent of initial parameter estimations, and converged successfully in each case tested.ConclusionThese investigations reveal that the iterative integral method is beneficial with respect to computing time, operator independence and robustness, and thus applicable at the bedside for this clinical application.

This manuscript presents the concerns around the increasingly common problem of not having readily available or useful “gold standard” measurements. This issue is particularly important in critical care where many measurements used in decision making are surrogates of what we would truly wish to use. However, the question is broad, important and applicable in many other areas.In particular, a gold standard measurement often exists, but is not clinically (or ethically in some cases) feasible. The question is how does one even begin to develop new measurements or surrogates if one has no gold standard to compare with?We raise this issue concisely with a specific example from mechanical ventilation, a core bread and butter therapy in critical care that is also a leading cause of length of stay and cost of care. Our proposed solution centers around a hierarchical validation approach that we believe would ameliorate ethics issues around radiation exposure that make current gold standard measures clinically infeasible, and thus provide a pathway to create a (new) gold standard.