Search

Modeling the respiratory system can aid clinical decision making during mechanical ventilation (MV) in the intensive care. However, most lung models only use airway pressure and flow data during inspiration phase of the breathing, and the data from expiration are often neglected. In this study, we hypothesized that the airway pressure and flow data during expiration have equal important information as the inspiratory data for respiratory mechanics modeling. In particular, expiratory time constant parameter, K, can provide unique information and relation with respiratory system elastance in MV patients. Two different respiratory models were investigated, with one model focusing on data from inspiratory cycle (Single compartment model) and the other focusing on expiratory cycle (Time constant model). The expiratory time constant model and single compartment model were evaluated based on 22 retrospective datasets of acute respiratory distress syndrome (ARDS) patients from two different cohorts. The expiration time constant model captured a moderate correlation with the lung elastance with R2= 0.568 in Cohort 1 and weak correlation with R2= 0.184 in Cohort 2, resulting in an overall of R2 = 0.435. Significant variations in lung resistance may lead to a poor correlation between Kand lung elastance, Elung. Thus, the application of Kas a surrogate to the respiratory elastance warrants further investigation.

Endogenous insulin secretion (UN) plays the leading role in glucose homeostasis. Understanding pathological changes in UNmay enable greater insight into the etiology of metabolic disorders particularly those related to hyperglycemia. The dynamic insulin sensitivity and secretion test (DISST) is a dynamic test that is able to quantify patient-specific insulin sensitivity (SI) values and UNprofiles. The DISST uses measured glucose, insulin and C-peptide assays with pharmaco-kinetic/dynamic glucose, insulin and C-peptide models to identify SIand UNprofiles. This study proposes a range of proportional-integral-derivative (PID) and proportional-derivative (PD) models to define UNas a function of glucose concentration. With relatively low percentage of residual error between measured C-peptide and fitted C-peptide response from the PID and PD models, it elucidates a more direct physiological link between insulin secretion to glucose concentration level.