The dimensional reduction method for identification of parameters that trade-off due to similar model roles.

Parameter identification is an important and widely used process across the field of biomedical engineering. However, it is susceptible to a number of potential difficulties, such as parameter trade-off, causing premature convergence at non-optimal parameter values. The proposed Dimensional Reduction Method (DRM) addresses this issue by iteratively reducing the dimension of hyperplanes where trade off occurs, and running subsequent identification processes within these hyperplanes. The DRM was validated using clinical data to optimize 4 parameters of the widely used Bergman Minimal Model of glucose and insulin kinetics, as well as in-silico data to optimize 5 parameters of the Pulmonary Recruitment (PR) Model. Results were compared with the popular Levenberg-Marquardt (LMQ) Algorithm using a Monte-Carlo methodology, with both methods afforded equivalent computational resources. The DRM converged to a lower or equal residual value in all tests run using the Bergman Minimal Model and actual patient data. For the PR model, the DRM attained significantly lower overall median parameter error values and lower residuals in the vast majority of tests. This shows the DRM has potential to provide better resolution of optimum parameter values for the variety of biomedical models in which significant levels of parameter trade-off occur.

Lung injury and respiratory mechanics in rugby union.

BACKGROUND: Rugby is a highly popular team contact sport associated with high injury rates. Specifically, there is a chance of inducing internal lung injuries as a result of the physical nature of the game. Such injuries are only identified with the use of specific invasive protocols or equipment. This study presents a model-based method to assess respiratory mechanics of N=11 rugby players that underwent a low intensity experimental Mechanical Ventilation (MV) Test before and after a rugby game. METHODS: Participants were connected to a ventilator via a facemask and their respiratory mechanics estimated using a time-varying elastance model. RESULTS: All participants had a respiratory elastance <10 cmH2O/L with no significant difference observed between pre and postgame respiratory mechanics (P>0.05). Model-based respiratory mechanics estimation has been used widely in the treatment of the critically ill in intensive care. However, the application of a ventilator to assess the respiratory mechanics of healthy human beings is limited. CONCLUSIONS: This method adapted from ICU mechanical ventilation can be used to provide insight to respiratory mechanics of healthy participants that can be used as a more precise measure of lung inflammation/injury that avoids invasive procedures. This is the first study to conceptualize the assessment of respiratory mechanics in healthy athletes as a means to monitor postexercise stress and therefore manage recovery.

An Assessment of the relationship between dicrotic notch timing and cardiac preload.

Cardiovascular disease (CVD) patient outcomes can be improved by extracting and synthesizing as much useful information as possible from a limited number of available measurements. An important metric in assessing the pathological state of CVD patients is cardiac preload. Left ventricular preload can be estimated through the surrogate measurement of left ventricular end diastolic volume (LVEDV). However, cardiac volumes are difficult to measure, clinically. This study develops a 3 parameter model relating the location of the dicrotic notch in the aortic waveform to LVEDV. This model was constructed using data from porcine experiments (N = 5 pietrain pigs, weights 20-28kg). The median difference between the observed LVEDV and modelled LVEDV was 5.4%, with a 100% range of 3.0% to 15.1%. Model parameters varied between individuals as well as contractile states. The median correlation was ρ = 0.77, with a minimum of 0.58 and maximum of 0.95. This model could be used to estimate prseload from the commonly measured aortic pressure waveform.

Modelling of the nonlinear end-systolic pressure-volume relation and volume-at-zero-pressure in porcine experiments.

The End-Systolic Pressure-Volume Relation (ESPVR) is generally modelled as a linear relationship between P and V as cardiac reflexes, such as the baroreflex, are typically suppressed in experiments. However, ESPVR has been observed to behave in a curvilinear fashion when cardiac reflexes are not suppressed, suggesting the curvilinear function may be more clinically appropriate. Data was gathered from 41 vena cava occlusion manoeuvres performed experimentally at a variety of PEEPs across 6 porcine specimens, and ESPVR determined for each pig. An exponential model of ESPVR was found to provide a higher correlation coefficient than a linear model in 6 out of 7 cases, and a lower Akaike Information Criterion (AIC) value in all cases. Further, the exponential ESPVR provided positive V0 values in a physiological range in 6 out of 7 cases analysed, while the linear ESPVR produced positive V0 values in only 3 out of 7 cases, suggesting linear extrapolation of ESPVR to determine V0 may be flawed.

Use of the DISST model to estimate the HOMA and Matsuda indexes using only a basal insulin assay.

It is hypothesized that early detection of reduced insulin sensitivity (SI) could prompt intervention that may reduce the considerable financial strain type 2 diabetes mellitus (T2DM) places on global health care. Reduction of the cost of already inexpensive SI metrics such as the Matsuda and HOMA indexes would enable more widespread, economically feasible use of these metrics for screening. The goal of this research was to determine a means of reducing the number of insulin samples and therefore the cost required to provide an accurate Matsuda Index value. The Dynamic Insulin Sensitivity and Secretion Test (DISST) model was used with the glucose and basal insulin measurements from an Oral Glucose Tolerance Test (OGTT) to predict patient insulin responses. The insulin response to the OGTT was determined via population based regression analysis that incorporated the 60-minute glucose and basal insulin values. The proposed method derived accurate and precise Matsuda Indices as compared to the fully sampled Matsuda (R = .95) using only the basal assay insulin-level data and 4 glucose measurements. Using a model employing the basal insulin also allows for determination of the 1-day HOMA value. The DISST model was successfully modified to allow for the accurate prediction an individual's insulin response to the OGTT. In turn, this enabled highly accurate and precise estimation of a Matsuda Index using only the glucose and basal insulin assays. As insulin assays account for the majority of the cost of the Matsuda Index, this model offers a significant reduction in assay cost.