Explore the vast range of titles available.

Overfitting and long training time are two fundamental challenges in multilayered neural network learning and deep learning in particular. Dropout and batch normalization are two well-recognized approaches to tackle these challenges. While both approaches share overlapping design principles, numerous research results have shown that they have unique strengths to improve deep learning. Many tools simplify these two approaches as a simple function call, allowing flexible stacking to form deep learning architectures. Although their usage guidelines are available, unfortunately no well-defined set of rules or comprehensive studies to investigate them concerning data input, network configurations, learning efficiency, and accuracy. It is not clear when users should consider using dropout and/or batch normalization, and how they should be combined (or used alternatively) to achieve optimized deep learning outcomes. In this paper we conduct an empirical study to investigate the effect of dropout and batch normalization on training deep learning models. We use multilayered dense neural networks and convolutional neural networks (CNN) as the deep learning models, and mix dropout and batch normalization to design different architectures and subsequently observe their performance in terms of training and test CPU time, number of parameters in the model (as a proxy for model size), and classification accuracy. The interplay between network structures, dropout, and batch normalization, allow us to conclude when and how dropout and batch normalization should be considered in deep learning. The empirical study quantified the increase in training time when dropout and batch normalization are used, as well as the increase in prediction time (important for constrained environments, such as smartphones and low-powered IoT devices). It showed that a non-adaptive optimizer (e.g. SGD) can outperform adaptive optimizers, but only at the cost of a significant amount of training times to perform hyperparameter tuning, while an adaptive optimizer (e.g. RMSProp) performs well without much tuning. Finally, it showed that dropout and batch normalization should be used in CNNs only with caution and experimentation (when in doubt and short on time to experiment, use only batch normalization).

Telehealth adoption accelerated in the past few years. Telehealth can be offered via simple video consultations or in more complex environments with doctors remotely assisting an operation or remote intensive care unit (ICU) support through a video feed. In these complex settings, it is essential for the doctors joining remotely to understand the remote environment quickly. The most critical information is identifying the patient’s location, which may not be immediately apparent in an operating room or ICU busy with equipment and other clinicians. In this paper, we trained object detection models to find the bounding box of patients in such complex hospital environments using a relatively small number of images (for an object detection task) from standard video cameras without depth or other information beyond a regular image. The best model achieves a high precision-recall area under the curve for this task, even when trained with a small dataset. We describe the process to identify the best model, considerations to deploy the model, and suggest improvements for future work.

Artificial intelligence (AI) had a few false starts – the AI winters of the 1970s and 1980s. We are now in what looks like an AI summer. There are many useful applications of AI in the field. But there are still unfulfilled promises and outright failures. From self-driving cars that work only in constrained cases, to medical image analysis products that would replace radiologists but never did, we still struggle to translate successful research into successful real-world applications.The software engineering community has accumulated a large body of knowledge over the decades on how to develop, release, and maintain products. AI products, being software products, benefit from some of that accumulated knowledge, but not all of it. AI products diverge from traditional software products in fundamental ways: their main component is not a specific piece of code, written for a specific purpose, but a generic piece of code, a model, customized by a training process driven by hyperparameters and a dataset. Datasets are usually large and models are opaque. We cannot directly inspect them as we can inspect the code of traditional software products. We need other methods to detect failures in AI products.This thesis investigates recent advancements that promote auditing, supported by transparency, as a mechanism to detect potential failures in AI products, focusing on healthcare applications. It reviews practices that apply to the early stages of the machine learning (ML) lifecycle, when datasets and models are created. These stages are unique to AI products.The proposed solution uses checklists to increase the transparency of these early stages. The thesis investigates checklists and guidelines that have been recently proposed in the AI industry and research communities, focusing on medical applications. Out of those checklists, it selected datasheets for dataset and model cards to increase the transparency of the dataset and model creation, respectively.As a demonstration of the increased transparency afforded by these methods, this thesis applies the selected checklists to a well-known medical imaging dataset, ChestX-ray8, and to a well-known model, CheXNet. The dataset datasheet created for ChestX-ray8 and the model card created for CheXNet show how a well-structured format increases transparency. The increased transparency, in turn, allows the ML community to interact with other stakeholders, e.g. domain experts in medical imaging, early on to find potential problems and opportunities for improvement.

Billions of X-ray images are taken worldwide each year. Machine learning, and deep learning in particular, has shown potential to help radiologists triage and diagnose images. However, deep learning requires large datasets with reliable labels. The CheXpert dataset was created with the participation of board-certified radiologists, resulting in the strong ground truth needed to train deep learning networks. Following the structured format of Datasheets for Datasets, this paper expands on the original CheXpert paper and other sources to show the critical role played by radiologists in the creation of reliable labels and to describe the different aspects of the dataset composition in detail. Such structured documentation intends to increase the awareness in the machine learning and medical communities of the strengths, applications, and evolution of CheXpert, thereby advancing the field of medical image analysis. Another objective of this paper is to put forward this dataset datasheet as an example to the community of how to create detailed and structured descriptions of datasets. We believe that clearly documenting the creation process, the contents, and applications of datasets accelerates the creation of useful and reliable models.

Acute liver failure (ALF) is a rapidly progressive heterogeneous illness with high mortality rate and no widely accessible cure. A promising drug candidate according to previous preclinical studies is the Reg3α (or HIP/PAP) lectin, which alleviates ALF through its free-radical scavenging activity. Here we study the therapeutic targets of Reg3α in order to gain information on the nature of the oxidative stress associated with ALF. Primary hepatocytes stressed with the reactive oxygen species (ROS) inducers TNFα and H2O2 were incubated with a recombinant Reg3α protein. ALF was induced in C57BL/6J mice by an anti-CD95 antibody. Livers and primary hepatocytes were harvested for deoxycholate separation of cellular and extracellular fractions, immunostaining, immunoprecipitation and malondialdehyde assays. Fibrin deposition was studied by immunofluorescence in frozen liver explants from patients with ALF. Fibrin deposition occurs during experimental and clinical acute liver injuries. Reg3α bound the resulting transient fibrin network, accumulated in the inflammatory extracellular matrix (ECM), greatly reduced extracellular ROS levels, and improved cell viability. Hepatocyte treatment with ligands of death receptors, e.g. TNFα and Fas, resulted in a twofold increase of malondialdehyde (MDA) level in the deoxycholate-insoluble fractions. Reg3α treatment maintained MDA at a level similar to control cells and thereby increased hepatocyte survival by 35%. No antioxidant effect of Reg3α was noted in the deoxycholate-soluble fractions. Preventing fibrin network formation with heparin suppressed the prosurvival effect of Reg3α. Reg3α is an ECM-targeted ROS scavenger that binds the fibrin scaffold resulting from hepatocyte death during ALF. ECM alteration is an important pathogenic factor of ALF and a relevant target for pharmacotherapy.

Desert dust transported from the Saharan-Sahel region to the Caribbean Sea is responsible for peak exposures of particulate matter (PM). This study explored the potential added value of satellite aerosol optical thickness (AOT) measurements, compared to the PM concentration at ground level, to retrospectively assess exposure during pregnancy. MAIAC MODIS AOT retrievals in blue band (AOT 470 ) were extracted for the French Guadeloupe archipelago. AOT 470 values and PM 10 concentrations were averaged over pregnancy for 906 women (2005–2008). Regression modeling was used to examine the AOT 470 -PM 10 relationship during pregnancy and test the association between dust exposure estimates and preterm birth. Moderate agreement was shown between mean AOT 470 retrievals and PM 10 ground-based measurements during pregnancy ( R 2  = 0.289). The magnitude of the association between desert dust exposure and preterm birth tended to be lower using the satellite method compared to the monitor method. The latter remains an acceptable trade-off between epidemiological relevance and exposure misclassification, in areas with few monitoring stations and complex topographical/meteorological conditions, such as tropical islands.

Introduction and hypothesis The objective was to describe the different laparoscopic and vaginal steps of sub-urethral infected mesh explantation as well as an unexpected and unusual complication: a sub-mucosal calcification on the sub-urethral segment of the sling that was not infiltrating the urethra. Methods This was carried out at our University Teaching Hospital of Strasbourg. Results We show the complete removal of an infected retropubic sling in a patient who had already undergone three previous surgeries without resolution of symptoms. This is a difficult case requiring a laparoscopic approach of the space of Retzius, which has been less familiar to surgeons since the advent of the midurethral sling. We show how to approach this space in an inflammatory environment by specifying its anatomical limits. Moreover, a great deal can be learned from the occurrence of an infectious complication after the surgery and the presence of a large calcification on the prosthesis. In this context, we advise a systematic antibiotic treatment to avoid this kind of complication. Conclusions Knowing the guidelines and the different surgical steps will help urogynecological surgeons to perform similar procedures in patients requiring removal of retropubic slings for complications such as infection and pain, where conservative management has not been successful. These cases must be discussed in a multidisciplinary meeting, as recommended by the French National Authority for Health, and managed in an expert establishment.

Women’s preoperative perceptions of pelvic-floor disorders may differ from those of their physicians. Our objective was to specify women’s hopes and fears before cystocele repair, and to compare them to those that surgeons anticipate. We performed a secondary qualitative analysis of data from the PROSPERE trial. Among the 265 women included, 98% reported at least one hope and 86% one fear before surgery. Sixteen surgeons also completed the free expectations-questionnaire as a typical patient would. Women’s hopes covered seven themes, and women’s fears eleven. Women’s hopes were concerning prolapse repair (60%), improvement of urinary function (39%), capacity for physical activities (28%), sexual function (27%), well-being (25%), and end of pain or heaviness (19%). Women’s fears were concerning prolapse relapse (38%), perioperative concerns (28%), urinary disorders (26%), pain (19%), sexual problems (10%), and physical impairment (6%). Surgeons anticipated typical hopes and fears which were very similar to those the majority of women reported. However, only 60% of the women reported prolapse repair as an expectation. Women’s expectations appear reasonable and consistent with the scientific literature on the improvement and the risk of relapse or complication related to cystocele repair. Our analysis encourages surgeons to consider individual woman’s expectations before pelvic-floor repair.

Acute liver failure (ALF) is a rapidly progressive heterogeneous illness with high mortality rate and no widely accessible cure. A promising drug candidate according to previous preclinical studies is the Reg3[alpha] (or HIP/PAP) lectin, which alleviates ALF through its free-radical scavenging activity. Here we study the therapeutic targets of Reg3[alpha] in order to gain information on the nature of the oxidative stress associated with ALF. Primary hepatocytes stressed with the reactive oxygen species (ROS) inducers TNF[alpha] and H.sub.2 O.sub.2 were incubated with a recombinant Reg3[alpha] protein. ALF was induced in C57BL/6J mice by an anti-CD95 antibody. Livers and primary hepatocytes were harvested for deoxycholate separation of cellular and extracellular fractions, immunostaining, immunoprecipitation and malondialdehyde assays. Fibrin deposition was studied by immunofluorescence in frozen liver explants from patients with ALF. Fibrin deposition occurs during experimental and clinical acute liver injuries. Reg3[alpha] bound the resulting transient fibrin network, accumulated in the inflammatory extracellular matrix (ECM), greatly reduced extracellular ROS levels, and improved cell viability. Hepatocyte treatment with ligands of death receptors, e.g. TNF[alpha] and Fas, resulted in a twofold increase of malondialdehyde (MDA) level in the deoxycholate-insoluble fractions. Reg3[alpha] treatment maintained MDA at a level similar to control cells and thereby increased hepatocyte survival by 35%. No antioxidant effect of Reg3[alpha] was noted in the deoxycholate-soluble fractions. Preventing fibrin network formation with heparin suppressed the prosurvival effect of Reg3[alpha]. Reg3[alpha] is an ECM-targeted ROS scavenger that binds the fibrin scaffold resulting from hepatocyte death during ALF. ECM alteration is an important pathogenic factor of ALF and a relevant target for pharmacotherapy.

Background and Aims Acute liver failure (ALF) is a rapidly progressive heterogeneous illness with high mortality rate and no widely accessible cure. A promising drug candidate according to previous preclinical studies is the Reg3[alpha] (or HIP/PAP) lectin, which alleviates ALF through its free-radical scavenging activity. Here we study the therapeutic targets of Reg3[alpha] in order to gain information on the nature of the oxidative stress associated with ALF. Methods Primary hepatocytes stressed with the reactive oxygen species (ROS) inducers TNF[alpha] and H.sub.2 O.sub.2 were incubated with a recombinant Reg3[alpha] protein. ALF was induced in C57BL/6J mice by an anti-CD95 antibody. Livers and primary hepatocytes were harvested for deoxycholate separation of cellular and extracellular fractions, immunostaining, immunoprecipitation and malondialdehyde assays. Fibrin deposition was studied by immunofluorescence in frozen liver explants from patients with ALF. Results Fibrin deposition occurs during experimental and clinical acute liver injuries. Reg3[alpha] bound the resulting transient fibrin network, accumulated in the inflammatory extracellular matrix (ECM), greatly reduced extracellular ROS levels, and improved cell viability. Hepatocyte treatment with ligands of death receptors, e.g. TNF[alpha] and Fas, resulted in a twofold increase of malondialdehyde (MDA) level in the deoxycholate-insoluble fractions. Reg3[alpha] treatment maintained MDA at a level similar to control cells and thereby increased hepatocyte survival by 35%. No antioxidant effect of Reg3[alpha] was noted in the deoxycholate-soluble fractions. Preventing fibrin network formation with heparin suppressed the prosurvival effect of Reg3[alpha]. Conclusions Reg3[alpha] is an ECM-targeted ROS scavenger that binds the fibrin scaffold resulting from hepatocyte death during ALF. ECM alteration is an important pathogenic factor of ALF and a relevant target for pharmacotherapy.