Comparing Amyloid Imaging Normalization Strategies for Alzheimer's Disease Classification using an Automated Machine Learning Pipeline.

Amyloid imaging has been widely used in Alzheimer's disease (AD) diagnosis and biomarker discovery through detecting the regional amyloid plaque density. It is essential to be normalized by a reference region to reduce noise and artifacts. To explore an optimal normalization strategy, we employ an automated machine learning (AutoML) pipeline, STREAMLINE, to conduct the AD diagnosis binary classification and perform permutation-based feature importance analysis with thirteen machine learning models. In this work, we perform a comparative study to evaluate the prediction performance and biomarker discovery capability of three amyloid imaging measures, including one original measure and two normalized measures using two reference regions (i.e., the whole cerebellum and the composite reference region). Our AutoML results indicate that the composite reference region normalization dataset yields a higher balanced accuracy, and identifies more AD-related regions based on the fractioned feature importance ranking.

Exploring Automated Machine Learning for Cognitive Outcome Prediction from Multimodal Brain Imaging using STREAMLINE.

STREAMLINE is a simple, transparent, end-to-end automated machine learning (AutoML) pipeline for easily conducting rigorous machine learning (ML) modeling and analysis. The initial version is limited to binary classification. In this work, we extend STREAMLINE through implementing multiple regression-based ML models, including linear regression, elastic net, group lasso, and L21 norm. We demonstrate the effectiveness of the regression version of STREAMLINE by applying it to the prediction of Alzheimer's disease (AD) cognitive outcomes using multimodal brain imaging data. Our empirical results demonstrate the feasibility and effectiveness of the newly expanded STREAMLINE as an AutoML pipeline for evaluating AD regression models, and for discovering multimodal imaging biomarkers.

Unsupervised modeling and genome-wide association identify novel features of allergic march trajectories.

BACKGROUND: The allergic march refers to the natural history of allergic conditions during infancy and childhood. However, population-level disease incidence patterns do not necessarily reflect the development of allergic disease in individuals. A better understanding of the factors that predispose to different allergic trajectories is needed. OBJECTIVE: Our aim was to determine the demographic and genetic features that are associated with the major allergic march trajectories. METHODS: Presence or absence of common allergic conditions (atopic dermatitis [AD], IgE-mediated food allergy [IgE-FA], asthma, and allergic rhinitis [AR]) was ascertained in a pediatric primary care birth cohort of 158,510 subjects. Hierarchic clustering and decision tree modeling were used to associate demographic features with allergic outcomes. Genome-wide association study was used to test for risk loci associated with specific allergic trajectories. RESULTS: We found an association between self-identified black race and progression from AD to asthma. Conversely, Asian or Pacific Islander race was associated with progression from AD to IgE-mediated food allergy, and white race was associated with progression from AD to AR. Genome-wide association study of trajectory groups identified risk loci associated with progression from AD to asthma (rs60242841) and from AD to AR (rs9565267, rs151041509, and rs78171803). Consistent with our epidemiologic associations, rs60242841 was more common in individuals of African ancestry than in individuals of European ancestry, whereas rs9565267 and rs151041509 were more common in individuals of European ancestry than in individuals of African ancestry. CONCLUSION: We have identified novel associations between race and progression along distinct allergic trajectories. Ancestral genetic differences may contribute to these associations. These results uncover important health disparities, refine the concept of the allergic march, and represent a step toward developing individualized medical approaches for these conditions.

Learning feature spaces for regression with genetic programming.

Genetic programming has found recent success as a tool for learning sets of features for regression and classification. Multidimensional genetic programming is a useful variant of genetic programming for this task because it represents candidate solutions as sets of programs. These sets of programs expose additional information that can be exploited for building block identification. In this work, we discuss this architecture and others in terms of their propensity for allowing heuristic search to utilize information during the evolutionary process. We investigate methods for biasing the components of programs that are promoted in order to guide search towards useful and complementary feature spaces. We study two main approaches: 1) the introduction of new objectives and 2) the use of specialized semantic variation operators. We find that a semantic crossover operator based on stagewise regression leads to significant improvements on a set of regression problems. The inclusion of semantic crossover produces state-of-the-art results in a large benchmark study of open-source regression problems in comparison to several state-of-the-art machine learning approaches and other genetic programming frameworks. Finally, we look at the collinearity and complexity of the data representations produced by different methods, in order to assess whether relevant, concise, and independent factors of variation can be produced in application.

Diagnostic biomarkers to differentiate sepsis from cytokine release syndrome in critically ill children.

Chimeric antigen receptor (CAR) T-cells directed against CD19 have drastically altered outcomes for children with relapsed and refractory acute lymphoblastic leukemia (r/r ALL). Pediatric patients with r/r ALL treated with CAR-T are at increased risk of both cytokine release syndrome (CRS) and sepsis. We sought to investigate the biologic differences between CRS and sepsis and to develop predictive models which could accurately differentiate CRS from sepsis at the time of critical illness. We identified 23 different cytokines that were significantly different between patients with sepsis and CRS. Using elastic net prediction modeling and tree classification, we identified cytokines that were able to classify subjects as having CRS or sepsis accurately. A markedly elevated interferon γ (IFNγ) or a mildly elevated IFNγ in combination with a low IL1ß were associated with CRS. A normal to mildly elevated IFNγ in combination with an elevated IL1ß was associated with sepsis. This combination of IFNγ and IL1ß was able to categorize subjects as having CRS or sepsis with 97% accuracy. As CAR-T therapies become more common, these data provide important novel information to better manage potential associated toxicities.

Mining Regional Imaging Genetic Associations via Voxel-wise Enrichment Analysis.

Brain imaging genetics aims to reveal genetic effects on brain phenotypes, where most studies examine phenotypes defined on anatomical or functional regions of interest (ROIs) given their biologically meaningful annotation and modest dimensionality compared with voxel-wise approaches. Typical ROI-level measures used in these studies are summary statistics from voxel-wise measures in the region, without making full use of individual voxel signals. In this paper, we propose a flexible and powerful framework for mining regional imaging genetic associations via voxel-wise enrichment analysis, which embraces the collective effect of weak voxel-level signals within an ROI. We demonstrate our method on an imaging genetic analysis using data from the Alzheimers Disease Neuroimaging Initiative, where we assess the collective regional genetic effects of voxel-wise FDGPET measures between 116 ROIs and 19 AD candidate SNPs. Compared with traditional ROI-wise and voxel-wise approaches, our method identified 102 additional significant associations, some of which were further supported by evidences in brain tissue-specific expression analysis. This demonstrates the promise of the proposed method as a flexible and powerful framework for exploring imaging genetic effects on the brain.

Can Haptic Simulators Distinguish Expert Performance? A Case Study in Central Venous Catheterization in Surgical Education.

INTRODUCTION: High-tech simulators are gaining popularity in surgical training programs because of their potential for improving clinical outcomes. However, most simulators are static in nature and only represent a single anatomical patient configuration. The Dynamic Haptic Robotic Training (DHRT) system was developed to simulate these diverse patient anatomies during Central Venous Catheterization (CVC) training. This article explores the use of the DHRT system to evaluate objective metrics for CVC insertion by comparing the performance of experts and novices. METHODS: Eleven expert surgeons and 13 first-year surgical residents (novices) performed multiple needle insertion trials on the DHRT system. Differences between expert and novice performance on the following five metrics were assessed using a multivariate analysis of variance: path length, standard deviation of deviations (SDoD), average velocity, distance to the center of the vessel, and time to complete (TtC) the needle insertion. A regression analysis was performed to identify if expertise could be predicted using these metrics. Then, a curve fit was conducted to identify whether learning curves were present for experts or novices on any of these five metrics. RESULTS: Time to complete the insertion and SDoD of the needle tip from an ideal path were significantly different between experts and novices. Learning curves were not present for experts but indicated a significant decrease in path length and TtC for novices. CONCLUSIONS: The DHRT system was able to identify significant differences in TtC and SDoD between experts and novices during CVC needle insertion procedures. In addition, novices were shown to improve their skills through DHRT training.

Semantic variation operators for multidimensional genetic programming.

Multidimensional genetic programming represents candidate solutions as sets of programs, and thereby provides an interesting framework for exploiting building block identification. Towards this goal, we investigate the use of machine learning as a way to bias which components of programs are promoted, and propose two semantic operators to choose where useful building blocks are placed during crossover. A forward stagewise crossover operator we propose leads to significant improvements on a set of regression problems, and produces state-of-the-art results in a large benchmark study. We discuss this architecture and others in terms of their propensity for allowing heuristic search to utilize information during the evolutionary process. Finally, we look at the collinearity and complexity of the data representations that result from these architectures, with a view towards disentangling factors of variation in application.

Building Ultrasound Phantoms With Modified Polyvinyl Chloride: A Comparison of Needle Insertion Forces and Sonographic Appearance With Commercial and Traditional Simulation Materials.

INTRODUCTION: Training using ultrasound phantoms allows for safe introduction to clinical skills and is associated with improved in-hospital performance. Many materials have been used to simulate human tissue in phantoms including commercial manikins, agar, gelatin, and Ballistics Gel; however, phantom tissues could be improved to provide higher-fidelity ultrasound images or tactile sensation. This article describes a novel phantom tissue mixture of a modified polyvinyl chloride (PVC) polymer, mineral oil, and chalk powder and evaluates needle cutting and ultrasonic properties of the modified PVC polymer mixture compared with a variety of phantom tissues. METHODS: The first experiment measured axial needle forces of a needle insertion into nine phantom materials, including three formulations of modified PVC. The second experiment used a pairwise comparison survey of ultrasound images to determine the perceived realism of phantom ultrasound images. RESULTS: It was found that the materials of Ballistics Gel and one of the PVC mixtures provide stiff force feedback similar to cadaver tissue. Other phantom materials including agar and gelatin provide very weak unrealistic force feedback. The survey results showed the PVC mixtures being viewed as the most realistic by the survey participants, whereas agar and Ballistics Gel were seen as the least realistic. CONCLUSIONS: The realism in cutting force and ultrasound visualization was determined for a variety of phantom materials. Novel modified PVC polymer has great potential for use in ultrasound phantoms because of its realistic ultrasound imaging and modifiable stiffness. This customizability allows for easy creation of multilayer tissue phantoms.

Critical Care Organizations: Building and Integrating Academic Programs.

OBJECTIVE: Academic medical centers in North America are expanding their missions from the traditional triad of patient care, research, and education to include the broader issue of healthcare delivery improvement. In recent years, integrated Critical Care Organizations have developed within academic centers to better meet the challenges of this broadening mission. The goal of this article was to provide interested administrators and intensivists with the proper resources, lines of communication, and organizational approach to accomplish integration and Critical Care Organization formation effectively. DESIGN: The Academic Critical Care Organization Building section workgroup of the taskforce established regular monthly conference calls to reach consensus on the development of a toolkit utilizing methods proven to advance the development of their own academic Critical Care Organizations. Relevant medical literature was reviewed by literature search. Materials from federal agencies and other national organizations were accessed through the Internet. SETTING: The Society of Critical Care Medicine convened a taskforce entitled "Academic Leaders in Critical Care Medicine" on February 22, 2016 at the 45th Critical Care Congress using the expertise of successful leaders of advanced governance Critical Care Organizations in North America to develop a toolkit for advancing Critical Care Organizations. MEASUREMENTS AND MAIN RESULTS: Key elements of an academic Critical Care Organization are outlined. The vital missions of multidisciplinary patient care, safety, and quality are linked to the research, education, and professional development missions that enhance the value of such organizations. Core features, benefits, barriers, and recommendations for integration of academic programs within Critical Care Organizations are described. Selected readings and resources to successfully implement the recommendations are provided. Communication with medical school and hospital leadership is discussed. CONCLUSIONS: We present the rationale for critical care programs to transition to integrated Critical Care Organizations within academic medical centers and provide recommendations and resources to facilitate this transition and foster Critical Care Organization effectiveness and future success.

Personalized Learning in Medical Education: Designing a User Interface for a Dynamic Haptic Robotic Trainer for Central Venous Catheterization.

While Virtual Reality (VR) has emerged as a viable method for training new medical residents, it has not yet reached all areas of training. One area lacking such development is surgical residency programs where there are large learning curves associated with skill development. In order to address this gap, a Dynamic Haptic Robotic Trainer (DHRT) was developed to help train surgical residents in the placement of ultrasound guided Internal Jugular Central Venous Catheters and to incorporate personalized learning. In order to accomplish this, a 2-part study was conducted to: (1) systematically analyze the feedback given to 18 third year medical students by trained professionals to identify the items necessary for a personalized learning system and (2) develop and experimentally test the usability of the personalized learning interface within the DHRT system. The results can be used to inform the design of VR and personalized learning systems within the medical community.

Considerations for higher efficiency and productivity in research activities.

There are several factors that are known to affect research productivity; some of them imply the need for large financial investments and others are related to work styles. There are some articles that provide suggestions for early career scientists (PhD students and postdocs) but few publications are oriented to professors about scientific leadership. As academic mentoring might be useful at all levels of experience, in this note we suggest several key considerations for higher efficiency and productivity in academic and research activities. More research is needed into the main work style features that differentiate highly productive scientists and research groups, as some of them could be innate and others could be transferable. As funding agencies, universities and research centers invest large amounts of money in order to have a better scientific productivity, a deeper understanding of these factors will be of high academic and societal impact.

IMPROVING MEDICAL EDUCATION: SIMULATING CHANGES IN PATIENT ANATOMY USING DYNAMIC HAPTIC FEEDBACK.

Virtual simulation is an emerging field in medical education. Research suggests that simulation reduces complication rates and improves learning gains for medical residents. One benefit of simulators is their allowance for more realistic and dynamic patient anatomies. While potentially useful throughout medical education, few studies have explored the impact of dynamic haptic simulators on medical training. In light of this research void, this study was developed to examine how a Dynamic-Haptic Robotic Trainer (DHRT) impacts medical student self-efficacy and skill gains compared to traditional simulators developed to train students in Internal Jugular Central Venous Catheter (IJ CVC) placement. The study was conducted with 18 third year medical students with no prior CVC insertion experience who underwent a pre-test, simulator training (manikin, robotic, or mixed) and post-test. The results revealed the DHRT as a useful method for training CVC skills and supports further research on dynamic haptic trainers in medical education.

Associations between Gut Microbial Colonization in Early Life and Respiratory Outcomes in Cystic Fibrosis.

OBJECTIVE: To examine patterns of microbial colonization of the respiratory and intestinal tracts in early life in infants with cystic fibrosis (CF) and their associations with breastfeeding and clinical outcomes. STUDY DESIGN: A comprehensive, prospective longitudinal analysis of the upper respiratory and intestinal microbiota in a cohort of infants and young children with CF followed from birth was performed. Genus-level microbial community composition was characterized using 16S-targeted pyrosequencing, and relationships with exposures and outcomes were assessed using linear mixed-effects models, time-to-event analysis, and principal components analysis. RESULTS: Sequencing of 120 samples from 13 subjects collected from birth to 34 months revealed relationships between breastfeeding, microbial diversity in the respiratory and intestinal tracts, and the timing of onset of respiratory complications, including exacerbations and colonization with Pseudomonas aeruginosa. Fluctuations in the abundance of specific bacterial taxa preceded clinical outcomes, including a significant decrease in bacteria of the genus Parabacteroides within the intestinal tract prior to the onset of chronic P aeruginosa colonization. Specific assemblages of bacteria in intestinal samples, but not respiratory samples, were associated with CF exacerbation in early life, indicating that the intestinal microbiome may play a role in lung health. CONCLUSIONS: Our findings relating breastfeeding to respiratory outcomes, gut diversity to prolonged periods of health, and specific bacterial communities in the gut prior to respiratory complications in CF highlight a connection between the intestinal microbiome and health and point to potential opportunities for antibiotic or probiotic interventions. Further studies in larger cohorts validating these findings are needed.

Effect of cardiopulmonary bypass on urea cycle intermediates and nitric oxide levels after congenital heart surgery.

OBJECTIVE: To test the hypothesis that cardiopulmonary bypass used for repair of ventricular septal defects and atrioventricular septal defects would decrease availability of urea cycle intermediates including arginine and subsequent nitric oxide availability. STUDY DESIGN: Consecutive infants (n = 26) undergoing cardiopulmonary bypass for repair of an unrestrictive ventricular septal defect or atrioventricular septal defect were studied. Blood samples were collected immediately before surgery, immediately after surgery, and 12 hours, 24 hours, and 48 hours after surgery. Urea cycle intermediates, including citrulline, arginine, and ornithine, were measured by amino acid analysis. Nitric oxide metabolites were measured by means of the modified Griess reaction. RESULTS: Cardiopulmonary bypass caused a significant decrease in the urea cycle intermediates arginine, citrulline, and ornithine at all postoperative time points compared with preoperative levels. The ratio of ornithine to citrulline, a marker of urea cycle function, was elevated at all postoperative time points compared with preoperative values, indicating decreased urea cycle function. Nitric oxide metabolites were significantly decreased at all postoperative time points except for 48 hours, compared with preoperative levels. CONCLUSIONS: Cardiopulmonary bypass significantly decreases availability of arginine, citrulline, and nitric oxide metabolites in the postoperative period. Decreased availability of nitric oxide precursors may contribute to the increased risk of postoperative pulmonary hypertension.