Rethinking Embedded Blocks for Machine Learning Applications

The underlying goal of FPGA architecture research is to devise flexible substrates that implement a wide variety of circuits efficiently. Contemporary FPGA architectures have been optimized to support networking, signal processing, and image processing applications through high-precision digital signal processing (DSP) blocks. The recent emergence of machine learning has created a new set of demands characterized by: (1) higher computational density and (2) low precision arithmetic requirements. With the goal of exploring this new design space in a methodical manner, we first propose a problem formulation involving computing nested loops over multiply-accumulate (MAC) operations, which covers many basic linear algebra primitives and standard deep neural network (DNN) kernels. A quantitative methodology for deriving efficient coarse-grained compute block architectures from benchmarks is then proposed together with a family of new embedded blocks, called MLBlocks. An MLBlock instance includes several multiply-accumulate units connected via a flexible routing, where each configuration performs a few parallel dot-products in a systolic array fashion. This architecture is parameterized with support for different data movements, reuse, and precisions, utilizing a columnar arrangement that is compatible with existing FPGA architectures. On synthetic benchmarks, we demonstrate that for 8-bit arithmetic, MLBlocks offer 6× improved performance over the commercial Xilinx DSP48E2 architecture with smaller area and delay; and for time-multiplexed 16-bit arithmetic, achieves 2× higher performance per area with the same area and frequency. All source codes and data, along with documents to reproduce all the results in this article, are available at http://github.com/raminrasoulinezhad/MLBlocks .

Roadrunner+: An Autonomous Intersection Management Cooperating with Connected Autonomous Vehicles and Pedestrians with Spillback Considered

The recent emergence of Connected Autonomous Vehicles (CAVs) enables the Autonomous Intersection Management (AIM) system, replacing traffic signals and human driving operations for improved safety and road efficiency. When CAVs approach an intersection, AIM schedules their intersection usage in a collision-free manner while minimizing their waiting times. In practice, however, there are pedestrian road-crossing requests and spillback problems, a blockage caused by the congestion of the downstream intersection when the traffic load exceeds the road capacity. As a result, collisions occur when CAVs ignore pedestrians or are forced to the congested road. In this article, we present a cooperative AIM system, named Roadrunner+ , which simultaneously considers CAVs, pedestrians, and upstream/downstream intersections for spillback handling, collision avoidance, and efficient CAV controls. The performance of Roadrunner+ is evaluated with the SUMO microscopic simulator. Our experimental results show that Roadrunner+ has 15.16% higher throughput than other AIM systems and 102.53% higher throughput than traditional traffic signals. Roadrunner+ also reduces 75.62% traveling delay compared to other AIM systems. Moreover, the results show that CAVs in Roadrunner+ save up to 7.64% in fuel consumption, and all the collisions caused by spillback are prevented in Roadrunner+.

Morpho-Functional Traits Reveal Differences in Size Fractionated Phytoplankton Communities but Do Not Significantly Affect Zooplankton Grazing

The recent emergence of approaches based on functional traits allows a more comprehensive evaluation of the role of functions and interactions within communities. As phytoplankton size and shape are the major determinants of its edibility to herbivores, alteration or loss of some morpho-functional phytoplankton traits should affect zooplankton grazing, fitness and population dynamics. Here, we investigated the response of altered phytoplankton morpho-functional trait distribution to grazing by zooplankton with contrasting food size preferences and feeding behaviors. To test this, we performed feeding trials in laboratory microcosms with size-fractionated freshwater phytoplankton (3 size classes, >30 µm; 5–30 µm and <5 µm) and two different consumer types: the cladoceran Daphnia longispina, (generalist unselective filter feeder) and the calanoid copepod Eudiaptomus sp. (selective feeder). We observed no significant changes in traits and composition between the controls and grazed phytoplankton communities. However, community composition and structure varied widely between the small and large size fractions, demonstrating the key role of size in structuring natural phytoplankton communities. Our findings also highlight the necessity to combine taxonomy and trait-based morpho-functional approaches when studying ecological dynamics in phytoplankton-zooplankton interactions.

A Vaccine Based on the A/ASIA/G-VII Lineage of Foot-and-Mouth Disease Virus Offers Low Levels of Protection against Circulating Viruses from the A/ASIA/Iran-05 lineage

The recent emergence and circulation of the A/ASIA/G-VII (A/G-VII) lineage of foot-and-mouth disease virus (FMDV) in the Middle East has resulted in the development of homologous vaccines to ensure susceptible animals are sufficiently protected against clinical disease. However, a second serotype A lineage called A/ASIA/Iran-05 (A/IRN/05) continues to circulate in the region and it is therefore imperative to ensure vaccine strains used will protect against both lineages. In addition, for FMDV vaccine banks that usually hold a limited number of strains, it is necessary to include strains with a broad antigenic coverage. To assess the cross protective ability of an A/G-VII emergency vaccine (formulated at 43 (95% CI 8–230) PD50/dose as determined during homologous challenge), we performed a heterologous potency test according to the European Pharmacopoeia design using a field isolate from the A/IRN/05 lineage as the challenge virus. The estimated heterologous potency in this study was 2.0 (95% CI 0.4–6.0) PD50/dose, which is below the minimum potency recommended by the World Organisation for Animal Health (OIE). Furthermore, the cross-reactive antibody titres against the heterologous challenge virus were poor (≤log10 0.9), even in those cattle that had received the full dose of vaccine. The geometric mean r1-value was 0.2 (95% CI 0.03–0.8), similar to the potency ratio of 0.04 (95% CI 0.004–0.3). Vaccination decreased viraemia and virus excretion compared to the unvaccinated controls. Our results indicate that this A/G-VII vaccine does not provide sufficient protection against viruses belonging to the A/IRN/05 lineage and therefore the A/G-VII vaccine strain cannot replace the A/IRN/05 vaccine strain but could be considered an additional strain for use in vaccines and antigen banks.

Modeling the recent outbreak of COVID-19 in India and its control strategies

The recent emergence of COVID-19 has drawn attention to the various methods of disease control. Since no proper treatment is available till date and the vaccination is restricted to certain age groups, also vaccine efficacy is still under progress, the emphasis has been given to the method of isolation and quarantine. This control is induced by tracing the contacts of the infectious individuals, putting them to the quarantine class and based on their symptoms, classifying them either as the susceptible or sick individuals and moving the sick individuals to the isolated class. To track the current pandemic situation of COVID-19 in India, we consider an extended Susceptible-Exposed-Quarantine-Infected-Isolated-Recovered (SEQ1IQ2R) compartmental model along with calculating its control reproductive number Rc. The disease can be kept in control if the value of Rc remains below one. This “threshold” value of Rc is used to optimize the period of quarantine, and isolation and have been calculated in order to eradicate the disease. The sensitivity analysis of Rc with respect to the quarantine and isolation period has also been done. Partial rank correlation coefficient method is applied to identify the most significant parameters involved in Rc. Based on the observed data, 7-days moving average curves are plotted for prelockdown, lockdown and unlock 1 phases. Following the trend of the curves for the infection, a generalized exponential function is used to estimate the data, and corresponding 95% confidence intervals are simulated to estimate the parameters. The effect of control measures such as quarantine and isolation are discussed. Following various mathematical and statistical tools, we systematically explore the impact of lockdown strategy in order to control the recent outbreak of COVID-19 transmission in India.

Integrating morphologic and molecular histopathological features through whole slide image registration and deep learning

Background Modern molecular pathology workflows in neuro-oncology heavily rely on the integration of morphologic and immunohistochemical patterns for analysis, classification, and prognostication. However, despite the recent emergence of digital pathology platforms and artificial intelligence-driven computational image analysis tools, automating the integration of histomorphologic information found across these multiple studies is challenged by large files sizes of whole slide images (WSIs) and shifts/rotations in tissue sections introduced during slide preparation. Methods To address this, we develop a workflow that couples different computer vision tools including scale-invariant feature transform (SIFT) and deep learning to efficiently align and integrate histopathological information found across multiple independent studies. We highlight the utility and automation potential of this workflow in the molecular subclassification and discovery of previously unappreciated spatial patterns in diffuse gliomas. Results First, we show how a SIFT-driven computer vision workflow was effective at automated WSI alignment in a cohort of 107 randomly selected surgical neuropathology cases (97/107 (91%) showing appropriate matches, AUC = 0.96). This alignment allows our AI-driven diagnostic workflow to not only differentiate different brain tumor types, but also integrate and carry out molecular subclassification of diffuse gliomas using relevant immunohistochemical biomarkers (IDH1-R132H, ATRX). To highlight the discovery potential of this workflow, we also examined spatial distributions of tumors showing heterogenous expression of the proliferation marker MIB1 and Olig2. This analysis helped uncovered an interesting and unappreciated association of Olig2 positive and proliferative areas in some gliomas (r = 0.62). Conclusion This efficient neuropathologist-inspired workflow provides a generalizable approach to help automate a variety of advanced immunohistochemically compatible diagnostic and discovery exercises in surgical neuropathology and neuro-oncology.

Unintended Side Effects of Electronic Cigarettes in Otolaryngology: A Scoping Review

Objective Electronic cigarettes (E-cigs) are nicotine delivery systems with increasing popularity. The US Food and Drug Administration defines side effects as unwanted or unexpected events or reactions. Our objective was to examine the unintended otolaryngology-related side effects associated with E-cigs. Data Sources Medline, EMBASE, CINAHL, Web of Science, and CENTRAL databases. Review Methods Study selection was independently performed by 2 authors in accordance with the PRISMA-ScR statement (Preferred Reporting Items for Systematic Reviews and Meta-analyses Extension for Scoping Reviews); discrepancies were resolved by the senior author. English studies from database inception to May 1, 2020, with a sample size >5 were included. In vitro, animal, and lower respiratory tract studies were excluded. The main outcome was defined as otolaryngology-related side effects following E-cig use. Levels of evidence per the Oxford Centre for Evidence-Based Medicine were used to determine study quality. Results From 1788 articles, 32 studies were included. The most common unintended side effects were throat irritation (n = 16), cough (n = 16), mouth irritation (n = 11), and oral mucosal lesions (n = 8). A large proportion of participants also reported conventional tobacco use in addition to E-cigs. Eight studies investigated the effectiveness of vaping on smoking cessation. The quality of the literature was level 2 to 4. Given the significant heterogeneity in the studies, meta-analysis was not performed. Conclusion The most reported side effects were throat and mouth irritation, followed by cough. The long-term impact of E-cigs is not known given the recent emergence of this technology. Future studies are warranted.

Image Guided Injection of Anterior Cruciate Ligament Tears with Autologous Bone Marrow Concentrate and Platelets: Midterm Analysis from A Randomized Controlled Trial

Background: There has been a recent emergence in the use of orthobiologics, including platelet-rich plasma (PRP) and bone marrow concentrate (BMC), in the treatment of various musculoskeletal conditions. The goal of this study was to determine if injection of BMC and platelet products into partial and full-thickness anterior cruciate ligament (ACL) tears can facilitate primary ligament healing in patients failing conservative care, resulting in improved outcomes compared to exercise therapy.Methods: Patients were randomized to either exercise therapy or percutaneous injection of autologous BMC with PRP and platelet lysate into the ACL under fluoroscopic guidance. Pain and function were assessed at baseline and at 1, 3, 6, 12, and 24 months. Baseline and 6-month post-treatment magnetic resonance imaging (MRI) were obtained to evaluate interval healing. Laxity was assessed using the Telos device.Results: There was significant improvement in functional outcomes in the BMC group, compared to base-line for LEFS at time points 3 up to 24 months s = 0.000000005), and significant improvement in pain in the BMC group at 6 (p = 0.00054), 12 (p = 0.00127), and 24 months (p = 0.002). There was no significant improvement in pain or function at any time point in the exercise therapy group. There was significant improvement in ACL MRI ImageJ quantitative assessment in the BMC group (p = 0.001) and no difference in the exercise group (p > 0.05). No serious adverse events were reported.Conclusion: Autologous BMC and platelet product injection into ACL tears improved patient function compared to exercise, observed through 24 months. Patients treated with BMC demonstrated quantitative improvements in post-treatment MRI scans suggestive of interval ligament healing.

Molecular detection of β-lactamase genes in Klebsiella pneumoniae and Escherichia coli isolated from different clinical sources

The rising occurrence of infections generated by Escherichia coli and Klebsiella pneumoniae that produce extended-spectrum β-lactamase (ESBL) is reason for concern. Due to the recent emergence of multidrug-resistant microorganisms that develop ESBL. The purpose of this work was to detect the ESBLs in clinical isolates of E. coli and K. pneumoniae. 118 samples of E. coli and 63 isolates of K. pneumoniae were collected from clinical samples. Polymerase chain reaction was used to detect β-lactamase genes (i.e., blaTEM, blaSHV, and blaCTX-M). Phenotypic detection revealed that 48.31% and 85.19% of E. coli and K. pneumoniae produced ESBLs, respectively. Whereas screening of ESBL genes in both bacteria employing a multiplex PCR test revealed that 24.58% of the ESBL-producing E. coli strains contained blaTEM, 50.85% contained blaSHV, and 32.2% contained blaCTX-M. Nevertheless, in K. pneumoniae, 40.74% blaTEM, 35.19% blaSHV, and 64.81% blaCTX-M genes were present. Antimicrobial resistance profiles of E. coli and K. pneumoniae isolates to twenty antibiotics were observed to vary significantly. Additionally, it was determined that the majority of E. coli and K. pneumoniae isolates were multidrug resistant (MDR). Additionally, 80.51% of E. coli isolates were resistant to the AMC antibiotic, while 0.00% were resistant to IPM and MEM. From the other hand, the resistant proportion of K. pneumoniae isolates was heterogeneous, ranging from 69.84% against CAZ to 0.00% against CIP and G antibiotics. The blaSHV gene was the most widespread among different forms of ESBLs in E. coli, but the most common gene in K. pneumoniae isolates was blaCTX-M (64.81%).

Antibody profiling reveals gender differences in response to SARS-COVID-2 infection

<abstract> <p>The recent emergence of the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has led to an ongoing global COVID-19 pandemic and public health crisis. Detailed study of human immune response to SARS-CoV-2 infection is the important topic for a successful treatment of this disease. Our study was aimed to characterize immune response on the level of antibody profiling in convalescent plasma of patients in Georgia. Antibodies against the following SARS-CoV-2 proteins were studied: nucleocapsid and various regions of spike (S) protein: S1, S2 and receptor binding domain (RBD). Convalescent plasma of patients 6–8 weeks after initial confirmation of SARS-CoV-2 infection were tested. Nearly 80% out of 162 patients studied showed presence of antibodies against nucleocapsid protein. The antibody response to three fragments of S protein was significantly less and varied in the range of 20–30%. Significantly more females as compared to males were producing antibodies against S1 fragment, whereas the difference between genders by the antibodies against nucleocapsid protein and RBD was statistically significant only by one-tailed Fisher exact test. There were no differences between the males and females by antibodies against S2 fragment. Thus, immune response against some viral antigens is stronger in females and we suggest that it could be one of the factors of less female fatality after SARS-CoV-2 infection.</p> </abstract>