An adversarial learning approach to generate pressure support ventilation waveforms for asynchrony detection.

BACKGROUND AND OBJECTIVE: Mechanical ventilation is a life-saving treatment for critically-ill patients. During treatment, patient-ventilator asynchrony (PVA) can occur, which can lead to pulmonary damage, complications, and higher mortality. While traditional detection methods for PVAs rely on visual inspection by clinicians, in recent years, machine learning models are being developed to detect PVAs automatically. However, training these models requires large labeled datasets, which are difficult to obtain, as labeling is a labour-intensive and time-consuming task, requiring clinical expertise. Simulating the lung-ventilator interactions has been proposed to obtain large labeled datasets to train machine learning classifiers. However, the obtained data lacks the influence of different hardware, of servo-controlled algorithms, and different sources of noise. Here, we propose VentGAN, an adversarial learning approach to improve simulated data by learning the ventilator fingerprints from unlabeled clinical data. METHODS: In VentGAN, the loss functions are designed to add characteristics of clinical waveforms to the generated results, while preserving the labels of the simulated waveforms. To validate VentGAN, we compare the performance for detection and classification of PVAs when training a previously developed machine learning algorithm with the original simulated data and with the data generated by VentGAN. Testing is performed on independent clinical data labeled by experts. The McNemar test is applied to evaluate statistical differences in the obtained classification accuracy. RESULTS: VentGAN significantly improves the classification accuracy for late cycling, early cycling and normal breaths (p< 0.01); no significant difference in accuracy was observed for delayed inspirations (p = 0.2), while the accuracy decreased for ineffective efforts (p< 0.01). CONCLUSIONS: Generation of realistic synthetic data with labels by the proposed framework is feasible and represents a promising avenue for improving training of machine learning models.

Evaluation of the accuracy of established patient inspiratory effort estimation methods during mechanical support ventilation.

There is a clinical need for monitoring inspiratory effort to prevent lung- and diaphragm injury in patients who receive supportive mechanical ventilation in an Intensive Care Unit. Different pressure-based techniques are available to estimate this inspiratory effort at the bedside, but the accuracy of their effort estimation is uncertain since they are all based on a simplified linear model of the respiratory system, which omits gas compressibility of air, and the viscoelasticity and nonlinearities of the respiratory system. The aim of this in-silico study was to provide an overview of the pressure-based estimation techniques and to evaluate their accuracy using a more sophisticated model of the respiratory system and ventilator. The influence of the following parameters on the accuracy of the pressure-based estimation techniques was evaluated using the in-silico model: 1) the patient's respiratory mechanics 2) PEEP and the inspiratory pressure of the ventilator 3) gas compressibility of air 4) viscoelasticity of the respiratory system 5) the strength of the inspiratory effort. The best-performing technique in terms of accuracy was the whole breath occlusion. The average error and maximum error were the lowest for all patient archetypes. We found that the error was related to the expansion of gas in the breathing set and lungs and respiratory compliance. However, concerns exist that other factors not included in the model, such as a changed muscle-force relation during an occlusion, might influence the true accuracy. The estimation techniques based on the esophageal pressure showed an error related to the viscoelastic element in the model which leads to a higher error than the occlusion. The error of the esophageal pressure-based techniques is therefore highly dependent on the pathology of the patient and the settings of the ventilator and might change over time while a patient recovers or becomes more ill.

A model-based approach to generating annotated pressure support waveforms.

Large numbers of asynchronies during pressure support ventilation cause discomfort and higher work of breathing in the patient, and are associated with an increased mortality. There is a need for real-time decision support to detect asynchronies and assist the clinician towards lung-protective ventilation. Machine learning techniques have been proposed to detect asynchronies, but they require large datasets with sufficient data diversity, sample size, and quality for training purposes. In this work, we propose a method for generating a large, realistic and labeled, synthetic dataset for training and validating machine learning algorithms to detect a wide variety of asynchrony types. We take a model-based approach in which we adapt a non-linear lung-airway model for use in a diverse patient group and add a first-order ventilator model to generate labeled pressure, flow, and volume waveforms of pressure support ventilation. The model was able to reproduce basic measured lung mechanics parameters. Experienced clinicians were not able to differentiate between the simulated waveforms and clinical data (P = 0.44 by Fisher's exact test). The detection performance of the machine learning trained on clinical data gave an overall comparable true positive rate on clinical data and on simulated data (an overall true positive rate of 94.3% and positive predictive value of 93.5% on simulated data and a true positive rate of 98% and positive predictive value of 98% on clinical data). Our findings demonstrate that it is possible to generate labeled pressure and flow waveforms with different types of asynchronies.

A Model-based Approach to Generating Annotated Pressure Support Waveforms.

During pressure support ventilation, every breath is triggered by the patient. Mismatches between the patient and the ventilator are called asynchronies. It has been reported that large numbers of asynchronies may be harmful and may lead to increased mortality. Automatic asynchrony detection and classification, with subsequent feedback to clinicians, will improve lung ventilation and, possibly, patient outcome. Machine learning techniques have been used to detect asynchronies. However, large, diverse and high-quality training and verification data sets are needed. In this work, we propose a model for generating a large, realistic, labeled, synthetic dataset for training and testing machine learning algorithms to detect a wide variety of asynchrony types. Next to a morphological evaluation of the obtained waveforms, validation of the proposed model includes a test with a machine learning algorithm trained on clinical data.

Comparison of Longitudinal Membrane Function in Peritoneal Dialysis Patients According to Dialysis Fluid Biocompatibility.

INTRODUCTION: Preservation of peritoneal function is essential in long-term peritoneal dialysis. Biocompatible dialysis solutions might prevent or postpone the membrane alteration resulting in ultrafiltration failure and consecutive morbidity and mortality. METHODS: We conducted an observational cohort study in which we made a longitudinal comparison between the course of peritoneal solute and fluid transport during treatment with conventional and biocompatible solutions. Therefore, prospectively collected peritoneal transport data from the yearly standard peritoneal permeability analysis were analyzed in 251 incident patients treated between 1994 and censoring in 2016. Fluid transport included small pore and free water transport. Solute transport was assessed by creatinine mass transfer area coefficient and glucose absorption. Linear mixed models including change point analyses were performed. Interaction with peritonitis was examined. RESULTS: One hundred thirty-five patients received conventional and 116 biocompatible solutions. Sixty-seven percent (conventional) and 64% (biocompatible) of these underwent minimally three transport measurements. Initially, biocompatible fluids showed higher small solute transport and lower ultrafiltration than conventional fluids up to 3 years. Thereafter, conventional fluids showed an increase in small solute transport (+2.7 ml/min per year; 95% confidence interval [CI]: 0.9 to 4.5) and a decrease of free water transport (-28.0 ml/min per year; 95% CI: -60.4 to 4.4). These were minor or absent in biocompatible treatment. Peritonitis induced a decrease of transcapillary ultrafiltration after 2 years on dialysis with conventional solutions (-291 ml/min per year; 95% CI: -550 to -32) while this was absent in biocompatible treatment. CONCLUSION: Despite a higher initial solute transport with biocompatible solutions, these have less influence on functional long-term peritoneal alterations than conventional solutions.

Olfactomedin 4 Serves as a Marker for Disease Severity in Pediatric Respiratory Syncytial Virus (RSV) Infection.

BACKGROUND: Respiratory viral infections follow an unpredictable clinical course in young children ranging from a common cold to respiratory failure. The transition from mild to severe disease occurs rapidly and is difficult to predict. The pathophysiology underlying disease severity has remained elusive. There is an urgent need to better understand the immune response in this disease to come up with biomarkers that may aid clinical decision making. METHODS: In a prospective study, flow cytometric and genome-wide gene expression analyses were performed on blood samples of 26 children with a diagnosis of severe, moderate or mild Respiratory Syncytial Virus (RSV) infection. Differentially expressed genes were validated using Q-PCR in a second cohort of 80 children during three consecutive winter seasons. FACS analyses were also performed in the second cohort and on recovery samples of severe cases in the first cohort. RESULTS: Severe RSV infection was associated with a transient but marked decrease in CD4+ T, CD8+ T, and NK cells in peripheral blood. Gene expression analyses in both cohorts identified Olfactomedin4 (OLFM4) as a fully discriminative marker between children with mild and severe RSV infection, giving a PAM cross-validation error of 0%. Patients with an OLFM4 gene expression level above -7.5 were 6 times more likely to develop severe disease, after correction for age at hospitalization and gestational age. CONCLUSION: By combining genome-wide expression profiling of blood cell subsets with clinically well-annotated samples, OLFM4 was identified as a biomarker for severity of pediatric RSV infection.

Treatment with anti-TNF alpha does not induce reactivation of latent Salmonella enterica serovar Typhimurium infection in C3H/HeN mice.

Therapy with tumour necrosis factor-alpha (TNFalpha)-blocking agents is successful in treating inflammatory disorders, but carries an increased risk of manifest and reactivating infection with intracellular bacteria. In a mouse model of latent Salmonella typhimurium infection, neutralization of TNFalpha did not result in reactivation of infection, suggesting only a minor role for TNFalpha during latency of persistent Salmonella infection.

Novel Salmonella enterica serovar Typhimurium protein that is indispensable for virulence and intracellular replication.

Upon contact with host cells, the intracellular pathogen Salmonella enterica serovar Typhimurium promotes its uptake, targeting, and survival in intracellular niches. In this process, the bacterium evades the microbicidal effector mechanisms of the macrophage, including oxygen intermediates. This study reports the phenotypic and genotypic characterization of an S. enterica serovar Typhimurium mutant that is hypersusceptible to superoxide. The susceptible phenotype is due to a MudJ insertion-inactivation of a previously undescribed Salmonella gene designated sspJ that is located between 54.4 and 64 min of the Salmonella chromosome and encodes a 392-amino-acid protein. In vivo, upon intraperitoneal injection of 10(4) to 10(7) bacteria in C3H/HeN and 10(1) to 10(4) bacteria in BALB/c mice, the mutant strain was less virulent than the wild type. Consistent with this finding, during the first hour after ingestion by macrophage-like J774 and RAW264.7 cells in vitro, the intracellular killing of the strain carrying sspJ::MudJ is enhanced fivefold over that of wild-type microorganisms. Wild-type salmonellae displayed significant intracellular replication during the first 24 h after uptake, but sspJ::MudJ mutants failed to do so. This phenotype could be restored to that of the wild type by sspJ complementation. The SspJ protein is found in the cytoplasmic membrane and periplasmic space. Amino acid sequence homology analysis did reveal a leader sequence and putative pyrroloquinoline quinone-binding domains, but no putative protein function. We excluded the possibility that SspJ is a scavenger of superoxide or has superoxide dismutase activity.

Cyclin D3 regulates proliferation and apoptosis of leukemic T cell lines.

Activation of the T cell receptor in leukemic T cell lines or T cell hybridomas causes growth inhibition. A similar growth inhibition is seen when protein kinase C is activated through addition of phorbol myristate acetate. This inhibition is due to an arrest of cell cycle progression in G(1) combined with an induction of apoptosis. Here we have investigated the mechanism by which these stimuli induce inhibition of proliferation in Jurkat and H9 leukemic T cell lines. We show that expression of cyclin D3 is reduced by each of these stimuli, resulting in a concomitant reduction in cyclin D-associated kinase activity. This reduction in cyclin D3-expression is crucial to the observed G(1) arrest, since ectopic expression of cyclin D3 can abrogate the G(1) arrest seen with each of these stimuli. Moreover, ectopic expression of cyclin D3 also prevents the induction of programmed cell death by phorbol myristate acetate and T-cell receptor activation, leading us to conclude that cyclin D3 not only plays a crucial role in progression through the G(1) phase, but is also involved in regulating apoptosis of T cells.

The influence of soil moisture content on the bioavailability and toxicity of cadmium for Folsomia candida Willem (Collembola: Isotomidae).

For Collembola living in the upper soil and litter layers, soil moisture is a primary factor for survival. In addition, a shortage or surplus of moisture might interact with other stress factors, such as persistent pollutants. The aim of this study therefore was to investigate the response of the collembolan species Folsomia candida to cadmium at different soil moisture contents. Tests were performed in an artificial soil substrate at soil moisture levels of 25, 35, 45, and 55%, corresponding with 74, 103, 132, and 162% of field capacity, respectively. Cadmium sorption to the soils, estimated from water-soluble concentrations, was not significantly affected by soil moisture content. Significant effects of soil moisture content were found for the toxicity of cadmium on body weight of F. candida after 4 and 6 weeks. EC50s did, however, not differ by more than a factor of 2. Reproduction in the controls was strongly affected by soil moisture content, but EC50s for the effect of cadmium on reproduction did not differ for the different moisture levels. Cadmium concentrations in the animals were not significantly affected by soil moisture content. From this study it can be concluded that, within the range chosen, soil moisture content does not have a great influence on the bioavailability and toxicity of cadmium for the collembolan F. candida.