Classification-based deep neural network vs mixture density network models for insulin sensitivity prediction problem

Model-based glycemic control (GC) protocols are used to treat stress-induced hyperglycaemia in intensive care units (ICUs). The STAR (Stochastic-TARgeted) glycemic control protocol - used in clinical practice in several ICUs in New Zealand, Hungary, Belgium, and Malaysia - is a model-based GC protocol using a patient-specific, model-based insulin sensitivity to describe the patient's actual state. Two neural network based methods are defined in this study to predict the patient's insulin sensitivity parameter: a classification deep neural network and a Mixture Density Network based method. Treatment data from three different patient cohorts are used to train the network models. Accuracy of neural network predictions are compared with the current model- based predictions used to guide care. The prediction accuracy was found to be the same or better than the reference. The authors suggest that these methods may be a promising alternative in model-based clinical treatment for patient state prediction. Still, more research is needed to validate these findings, including in-silico simulations and clinical validation trials.

Structural priors represented by discrete cosine transform improve EIT functional imaging

Structural prior information can improve electrical impedance tomography (EIT) reconstruction. In this contribution, we introduce a discrete cosine transformation-based (DCT-based) EIT reconstruction algorithm to demonstrate a way to incorporate the structural prior with the EIT reconstruction process. Structural prior information is obtained from other available imaging methods, e.g., thorax-CT. The DCT-based approach creates a functional EIT image of regional lung ventilation while preserving the introduced structural information. This leads to an easier interpretation in clinical settings while maintaining the advantages of EIT in terms of bedside monitoring during mechanical ventilation. Structural priors introduced in the DCT-based approach are of two categories in terms of different levels of information included: a contour prior only differentiates lung and non-lung region, while a detail prior includes information, such as atelectasis, within the lung area. To demonstrate the increased interpretability of the EIT image through structural prior in the DCT-based approach, the DCT-based reconstructions were compared with reconstructions from a widely applied one-step Gauss-Newton solver with background prior and from the advanced GREIT algorithm. The comparisons were conducted both on simulation data and retrospective patient data. In the simulation, we used two sets of forward models to simulate different lung conditions. A contour prior and a detail prior were derived from simulation ground truth. With these two structural priors, the reconstructions from the DCT-based approach were compared with the reconstructions from both the one-step Gauss-Newton solver and the GREIT. The difference between the reconstructions and the simulation ground truth is calculated by the ℓ2-norm image difference. In retrospective patient data analysis, datasets from six lung disease patients were included. For each patient, a detail prior was derived from the patient's CT, respectively. The detail prior was used for the reconstructions using the DCT-based approach, which was compared with the reconstructions from the GREIT. The reconstructions from the DCT-based approach are more comprehensive and interpretable in terms of preserving the structure specified by the priors, both in simulation and retrospective patient data analysis. In simulation analysis, the ℓ2-norm image difference of the DCT-based approach with a contour prior decreased on average by 34% from GREIT and 49% from the Gauss-Newton solver with background prior; for reconstructions of the DCT-based approach with detail prior, on average the ℓ2-norm image difference is 53% less than GREIT and 63% less than the reconstruction with background prior. In retrospective patient data analysis, the reconstructions from both the DCT-based approach and GREIT can indicate the current patient status, but the DCT-based approach yields more interpretable results. However, it is worth noting that the preserved structure in the DCT-based approach is derived from another imaging method, not from the EIT measurement. If the structural prior is outdated or wrong, the result might be misleadingly interpreted, which induces false clinical conclusions. Further research in terms of evaluating the validity of the structural prior and detecting the outdated prior is necessary.

Differentiating Phenotypes of Coronavirus Disease-2019 Pneumonia by Electric Impedance Tomography

Introduction

Coronavirus disease-2019 (COVID-19) pneumonia has different phenotypes. Selecting the patient individualized and optimal respirator settings for the ventilated patient is a challenging process. Electric impedance tomography (EIT) is a real-time, radiation-free functional imaging technique that can aid clinicians in differentiating the “low” (L-) and “high” (H-) phenotypes of COVID-19 pneumonia described previously.

Methods

Two patients (“A” and “B”) underwent a stepwise positive end-expiratory pressure (PEEP) recruitment by 3 cmH₂O of steps from PEEP 10 to 25 and back to 10 cmH₂O during a pressure control ventilation of 15 cmH₂O. Recruitment maneuvers were performed under continuous EIT recording on a daily basis until patients required controlled ventilation mode.

Results

Patients “A” and “B” had a 7- and 12-day long trial, respectively. At the daily baseline, patient “A” had significantly higher compliance: mean ± SD = 53 ± 7 vs. 38 ± 5 ml/cmH₂O (p < 0.001) and a significantly higher physiological dead space according to the Bohr–Enghoff equation than patient “B”: mean ± SD = 52 ± 4 vs. 45 ± 6% (p = 0.018). Following recruitment maneuvers, patient “A” had a significantly higher cumulative collapse ratio detected by EIT than patient “B”: mean ± SD = 0.40 ± 0.08 vs. 0.29 ± 0.08 (p = 0.007). In patient “A,” there was a significant linear regression between the cumulative collapse ratios at the end of the recruitment maneuvers (R² = 0.824, p = 0.005) by moving forward in days, while not for patient “B” (R² = 0.329, p = 0.5).

Conclusion

Patient “B” was recognized as H-phenotype with high elastance, low compliance, higher recruitability, and low ventilation-to-perfusion ratio; meanwhile patient “A” was identified as the L-phenotype with low elastance, high compliance, and lower recruitability. Observation by EIT was not just able to differentiate the two phenotypes, but it also could follow the transition from L- to H-type within patient “A.”

Clinical Trial Registration

www.ClinicalTrials.gov, identifier: NCT04360837.

Detection of Different COVID-19 Pneumonia Phenotypes with Estimated Alveolar Collapse and Overdistention by Bedside Electrical Impedance Tomography

COVID-19 induced acute respiratory distress syndrome (ARDS) could have two different phenotypes, which was reported to have different response and outcome to the typical ARDS positive end-expiration pressure (PEEP) treatment. The identification of the different phenotypes in terms of the recruitability can help improve the patient outcome. In this contribution we conducted alveolar overdistention and collapse analysis with the long term electrical impedance tomography monitoring data on two severe COVID-19 pneumonia patients. The result showed different patient reactions to the PEEP trial, revealed the progressive change in the patient status, and indicted a possible phenotype transition in one patient. It might suggest that EIT can be a practical tool to identify phenotypes and to provide progressive information of COVID-19 pneumonia.

Modelling intestinal glucose absorption in premature infants using continuous glucose monitoring data

BACKGROUND

Model-based glycaemic control protocols have shown promise in neonatal intensive care units (NICUs) for reducing both hyperglycaemia and insulin-therapy driven hypoglycaemia. However, current models for the appearance of glucose from enteral feeding are based on values from adult intensive care cohorts. This study aims to determine enteral glucose appearance model parameters more reflective of premature infant physiology.

METHODS

Peaks in CGM data associated with enteral milk feeds in preterm and term infants are used to fit a two compartment gut model. The first compartment describes glucose in the stomach, and the half life of gastric emptying is estimated as 20 min from literature. The second compartment describes glucose in the small intestine, and absorption of glucose into the blood is fit to CGM data. Two infant cohorts from two NICUs are used, and results are compared to appearances derived from data in highly controlled studies in literature.

RESULTS

The average half life across all infants for glucose absorption from the gut to the blood was 50 min. This result was slightly slower than, but of similar magnitude to, results derived from literature. No trends were found with gestational or postnatal age. Breast milk fed infants were found to have a higher absorption constant than formula fed infants, a result which may reflect known differences in gastric emptying for different feed types.

CONCLUSIONS

This paper presents a methodology for estimation of glucose appearance due to enteral feeding, and model parameters suitable for a NICU model-based glycaemic control context.

Safety, efficacy and clinical generalization of the STAR protocol: a retrospective analysis

Background

The changes in metabolic pathways and metabolites due to critical illness result in a highly complex and dynamic metabolic state, making safe, effective management of hyperglycemia and hypoglycemia difficult. In addition, clinical practices can vary significantly, thus making GC protocols difficult to generalize across units.The aim of this study was to provide a retrospective analysis of the safety, performance and workload of the stochastic targeted (STAR) glycemic control (GC) protocol to demonstrate that patient-specific, safe, effective GC is possible with the STAR protocol and that it is also generalizable across/over different units and clinical practices.

Methods

Retrospective analysis of STAR GC in the Christchurch Hospital Intensive Care Unit (ICU), New Zealand (267 patients), and the Gyula Hospital, Hungary (47 patients), is analyzed (2011–2015). STAR Christchurch (BG target 4.4–8.0 mmol/L) is also compared to the Specialized Relative Insulin and Nutrition Tables (SPRINT) protocol (BG target 4.4–6.1 mmol/L) implemented in the Christchurch Hospital ICU, New Zealand (292 patients, 2005–2007). Cohort mortality, effectiveness and safety of glycemic control and nutrition delivered are compared using nonparametric statistics.

Results

Both STAR implementations and SPRINT resulted in over 86 % of time per episode in the blood glucose (BG) band of 4.4–8.0 mmol/L. Patients treated using STAR in Christchurch ICU spent 36.7 % less time on protocol and were fed significantly more than those treated with SPRINT (73 vs. 86 % of caloric target). The results from STAR in both Christchurch and Gyula were very similar, with the BG distributions being almost identical. STAR provided safe GC with very few patients experiencing severe hypoglycemia (BG < 2.2 mmol/L, <5 patients, 1.5 %).

Conclusions

STAR outperformed its predecessor, SPRINT, by providing higher nutrition and equally safe, effective control for all the days of patient stay, while lowering the number of measurements and interventions required. The STAR protocol has the ability to deliver high performance and high safety across patient types, time, clinical practice culture (Christchurch and Gyula) and clinical resources.

Daily evolution of insulin sensitivity variability with respect to diagnosis in the critically ill

INTRODUCTION

This study examines the likelihood and evolution of overall and hypoglycemia-inducing variability of insulin sensitivity in ICU patients based on diagnosis and day of stay.

MATERIALS AND METHODS

An analysis of model-based insulin sensitivity for n=390 patients in a medical ICU (Christchurch, New Zealand). Two metrics are defined to measure the variability of a patient's insulin sensitivity relative to predictions of a stochastic model created from the same data for all patients over all days of stay. The first selectively captures large increases related to the risk of hypoglycemia. The second captures overall variability. Distributions of per-patient variability scores were evaluated over different ICU days of stay and for different diagnosis groups based on APACHE III: operative and non-operative cardiac, gastric, all other. Linear and generalized linear mixed effects models assess the statistical significance of differences between groups and over days.

RESULTS

Variability defined by the two metrics was not substantially different. Variability was highest on day 1, and decreased over time (p<0.0001) in every diagnosis group. There were significant differences between some diagnosis groups: non-operative gastric patients were the least variable, while cardiac (operative and non-operative) patients exhibited the highest variability.

CONCLUSIONS

This study characterizes the variability and evolution of insulin sensitivity in critically ill patients, and may help inform the clinical management of metabolic dysfunction in critical care.

Identification of dental root canals and their medial line from micro-CT and cone-beam CT records

Background

Shape of the dental root canal is highly patient specific. Automated identification methods of the medial line of dental root canals and the reproduction of their 3D shape can be beneficial for planning endodontic interventions as severely curved root canals or multi-rooted teeth may pose treatment challenges. Accurate shape information of the root canals may also be used by manufacturers of endodontic instruments in order to make more efficient clinical tools.

Method

Novel image processing procedures dedicated to the automated detection of the medial axis of the root canal from dental micro-CT and cone-beam CT records are developed. For micro-CT, the 3D model of the root canal is built up from several hundred parallel cross sections, using image enhancement, histogram based fuzzy c-means clustering, center point detection in the segmented slice, three dimensional inner surface reconstruction, and potential field driven curve skeleton extraction in three dimensions. Cone-beam CT records are processed with image enhancement filters and fuzzy chain based regional segmentation, followed by the reconstruction of the root canal surface and detecting its skeleton via a mesh contraction algorithm.

Results

The proposed medial line identification and root canal detection algorithms are validated on clinical data sets. 25 micro-CT and 36 cone-beam-CT records are used in the validation procedure. The overall success rate of the automatic dental root canal identification was about 92% in both procedures. The algorithms proved to be accurate enough for endodontic therapy planning.

Conclusions

Accurate medial line identification and shape detection algorithms of dental root canal have been developed. Different procedures are defined for micro-CT and cone-beam CT records. The automated execution of the subsequent processing steps allows easy application of the algorithms in the dental care. The output data of the image processing procedures is suitable for mathematical modeling of the central line. The proposed methods can help automate the preparation and design of several kinds of endodontic interventions.

Efficient inhomogeneity compensation using fuzzy c-means clustering models

Intensity inhomogeneity or intensity non-uniformity (INU) is an undesired phenomenon that represents the main obstacle for magnetic resonance (MR) image segmentation and registration methods. Various techniques have been proposed to eliminate or compensate the INU, most of which are embedded into classification or clustering algorithms, they generally have difficulties when INU reaches high amplitudes and usually suffer from high computational load. This study reformulates the design of c-means clustering based INU compensation techniques by identifying and separating those globally working computationally costly operations that can be applied to gray intensity levels instead of individual pixels. The theoretical assumptions are demonstrated using the fuzzy c-means algorithm, but the proposed modification is compatible with a various range of c-means clustering based INU compensation and MR image segmentation algorithms. Experiments carried out using synthetic phantoms and real MR images indicate that the proposed approach produces practically the same segmentation accuracy as the conventional formulation, but 20-30 times faster.

Towards unified electromagnetic tracking system assessment-static errors

Recent advances in Image-Guided Surgery allows physicians to incorporate up-to-date, high quality patient data in the surgical decision making, and sometimes to directly perform operations based on pre- or intra-operatively acquired patient images. Electromagnetic tracking is the fastest growing area within, where the position and orientation of tiny sensors can be determined with sub-millimeter accuracy in the field created by a generator. One of the major barriers to the wider spread of electromagnetic tracking solutions is their susceptibility to ferromagnetic materials and external electromagnetic sources. The research community has long been engaged with the topic to find engineering solutions to increase measurement reliability and accuracy. This article gives an overview of related experiments, and presents our recommendation towards a robust method to collect representative data about electromagnetic trackers.

Induced L₂-norm minimization of glucose-insulin system for Type I diabetic patients

Using induced L₂-norm minimization, a robust controller was developed for insulin delivery in Type I diabetic patients. The high-complexity nonlinear diabetic patient Sorensen-model was considered and Linear Parameter Varying methodology was used to develop open-loop model and robust H(∞) controller. Considering the normoglycaemic set point (81.1 mg/dL), a polytopic set was created over the physiologic boundaries of the glucose-insulin interaction of the Sorensen-model. In this way, Linear Parameter Varying model formalism was defined. The robust control was developed considering input and output multiplicative uncertainties with two additional uncertainties from those used in the literature: sensor noise and worst-case design for meal disturbance (60 g carbohydrate). Simulation scenario on large meal absorption illustrates the applicability of the robust LPV control technique, while patient variability is tested with real data taken from the SPRINT clinical protocol on ICU patients.

Hypersensitivity to thromboxane receptor mediated cerebral vasomotion and CBF oscillations during acute NO-deficiency in rats

BACKGROUND

Low frequency (4-12 cpm) spontaneous fluctuations of the cerebrovascular tone (vasomotion) and oscillations of the cerebral blood flow (CBF) have been reported in diseases associated with endothelial dysfunction. Since endothelium-derived nitric oxide (NO) suppresses constitutively the release and vascular effects of thromboxane A(2) (TXA(2)), NO-deficiency is often associated with activation of thromboxane receptors (TP). In the present study we hypothesized that in the absence of NO, overactivation of the TP-receptor mediated cerebrovascular signaling pathway contributes to the development of vasomotion and CBF oscillations.

METHODOLOGY/PRINCIPAL FINDINGS

Effects of pharmacological modulation of TP-receptor activation and its downstream signaling pathway have been investigated on CBF oscillations (measured by laser-Doppler flowmetry in anesthetized rats) and vasomotion (measured by isometric tension recording in isolated rat middle cerebral arteries, MCAs) both under physiological conditions and after acute inhibition of NO synthesis. Administration of the TP-receptor agonist U-46619 (1 µg/kg i.v.) to control animals failed to induce any changes of the systemic or cerebral circulatory parameters. Inhibition of the NO synthesis by nitro-L-arginine methyl ester (L-NAME, 100 mg/kg i.v.) resulted in increased mean arterial blood pressure and a decreased CBF accompanied by appearance of CBF-oscillations with a dominant frequency of 148±2 mHz. U-46619 significantly augmented the CBF-oscillations induced by L-NAME while inhibition of endogenous TXA(2) synthesis by ozagrel (10 mg/kg i.v.) attenuated it. In isolated MCAs U-46619 in a concentration of 100 nM, which induced weak and stable contraction under physiological conditions, evoked sustained vasomotion in the absence of NO, which effect could be completely reversed by inhibition of Rho-kinase by 10 µM Y-27632.

CONCLUSION/SIGNIFICANCE

These results suggest that hypersensitivity of the TP-receptor-Rho-kinase signaling pathway contributes to the development of low frequency cerebral vasomotion which may propagate to vasospasm in pathophysiological states associated with NO-deficiency.

Robust blood-glucose control using Mathematica

A robust control design on frequency domain using Mathematica is presented for regularization of glucose level in type I diabetes persons under intensive care. The method originally proposed under Mathematica by Helton and Merino, --now with an improved disturbance rejection constraint inequality--is employed, using a three-state minimal patient model. The robustness of the resulted high-order linear controller is demonstrated by nonlinear closed loop simulation in state-space, in case of standard meal disturbances and is compared with H infinity design implemented with the mu-toolbox of Matlab. The controller designed with model parameters represented the most favorable plant dynamics from the point of view of control purposes, can operate properly even in case of parameter values of the worst-case scenario.

Adaptation of the hypothalamic blood flow to chronic nitric oxide deficiency is independent of vasodilator prostanoids

The aim of our study was to investigate the adaptation of the hypothalamic circulation to chronic nitric oxide (NO) deficiency in rats. Hypothalamic blood flow (HBF) remained unaltered during chronic oral administration of the NO synthase (NOS) inhibitor N(G)-nitro-l-arginine methyl ester (l-NAME, 1 mg/ml drinking water) although acute NOS blockade by intravenous l-NAME injection (50 mg/kg) induced a dramatic HBF decrease. In chronically NOS blocked animals, however, acute l-NAME administration failed to influence the HBF. Reversal of chronic NOS blockade by intravenous l-arginine infusion evoked significant hypothalamic hyperemia suggesting the appearance of a compensatory vasodilator mechanism in the absence of NO. In order to clarify the potential involvement of vasodilator prostanoids in this adaptation, cyclooxygenase (COX) mRNA and protein levels were determined in the hypothalamus, but none of the known isoenzymes (COX-1, COX-2, COX-3) showed upregulation after chronic NOS blockade. Furthermore, levels of vasodilator prostanoid (PGI(2), PGE(2) and PGD(2)) metabolites were also not elevated. Interestingly, however, hypothalamic levels of vasoconstrictor prostanoids (TXA(2) and PGF(2alpha)) decreased after chronic NOS blockade. COX inhibition by indomethacin but not by diclofenac decreased the HBF in control animals. However, neither indomethacin nor diclofenac induced an altered HBF-response after chronic l-NAME treatment. Although urinary excretion of PGI(2) and PGE(2) metabolites markedly increased during chronic NOS blockade, indicating COX activation in the systemic circulation, we conclude that the adaptation of the hypothalamic circulation to the reduction of NO synthesis is independent of vasodilator prostanoids. Reduced release of vasoconstrictor prostanoids, however, may contribute to the normalization of HBF after chronic loss of NO.

Classification of Time Series Using Singular Values and Wavelet Subband Analysis with ANN and SVM Classifiers

Oscillation of cerebral blood flow (CBF) in physiological or pathophysiological brain states is common, therefore it is promising to identify cerebral circulation disorders based on CBF signal classification. To characterize temporal blood flow patterns, we applied two feature extractions, spectral matrix and wavelet subband analysis. To distinguish between different physiological states, two different classifications have been developed - the radial basis function-based neural network and a support vector classifier with a Gaussian kernel. Feature extraction and classification are evaluated and their efficiency compared. Calculation was done using <I>Mathematica</I> 5.1 and its <I>Wavelet Application</I>.

In search of the nature of specific nucleic acid-protein interactions

The theory of "codon-amino acid coevolution" was first proposed by Woese in 1967. It suggests that there is a stereochemical matching - that is, affinity - between amino acids and certain of the base triplet sequences that code for those amino acids. We have constructed a Common Periodic Table of Codons and Amino Acids, where the Nucleic Acid Table showed perfect axial symmetry for codons and the corresponding Amino Acid Table also displayed periodicity regarding the biochemical properties (charge and hydrophobicity) of the 20 amino acids and the position of the stop signals. The Table indicates that the middle (2nd) amino acid in the codon has a prominent role in determining some of the structural features of the amino acids. The possibility that physical contact between codons and amino acids might exist was tested on restriction enzymes. Many recognition site-like sequences were found in the coding sequences of these enzymes and as many as 73 examples of codon-amino acid co-location were observed in the 7 known 3D structures (December 2003) of endonuclease-nucleic acid complexes. These results indicate that the smallest possible units of specific nucleic acid-protein interaction are indeed the stereochemically compatible codons and amino acids.

A fully symbolic design and modeling of nonlinear glucose control with Control System Professional Suite (CSPS) of Mathematica

In this case study a fully symbolic design and modeling method are presented for blood glucose control of diabetic patients under intensive care using Mathematica. The analysis is based on a modified two-compartment model proposed by Bergman et al. The applied feedback control law decoupling even the nonlinear model leads to a fully symbolic solution of the closed loop equations. The effectivity of the applied symbolic procedures being mostly built-in the new version of Control System Professional Suite (CSPS) Application of Mathematica have been demonstrated for controller design in case of a glucose control for treatment of diabetes mellitus and also presented for a numerical situation described in Juhász. The results are in good agreement with the earlier presented symbolic-numeric analysis by Benyó et al.

A common periodic table of codons and amino acids

A periodic table of codons has been designed where the codons are in regular locations. The table has four fields (16 places in each) one with each of the four nucleotides (A, U, G, C) in the central codon position. Thus, AAA (lysine), UUU (phenylalanine), GGG (glycine), and CCC (proline) were placed into the corners of the fields as the main codons (and amino acids) of the fields. They were connected to each other by six axes. The resulting nucleic acid periodic table showed perfect axial symmetry for codons. The corresponding amino acid table also displaced periodicity regarding the biochemical properties (charge and hydropathy) of the 20 amino acids and the position of the stop signals. The table emphasizes the importance of the central nucleotide in the codons and predicts that purines control the charge while pyrimidines determine the polarity of the amino acids. This prediction was experimentally tested.

An open architecture patient monitoring system using standard technologies

Computer-aided bedside patient monitoring is applied in areas where real-time vital function analysis takes place. Modern bedside monitoring requires not only the networking of bedside monitors with a central monitor but also other standard communication interfaces. In this paper, a novel approach to patient monitoring is introduced. A patient monitoring system was developed and implemented based on an existing industry standard communication network, using standard hardware components and software technologies. The open architecture system design offers scalability, standard interfaces, and flexible signal interpretation possibilities.