Accurate and timely arrhythmia multi-classification is crucial for diagnosing and managing cardiovascular diseases. Traditional interpretation of electrocardiograms is timeconsuming and prone to human error. To address these challenges, this study describes a novel deep learning-based approach for automated arrhythmia classification. Our proposed model, a convolutional-recurrent neural network, leverages the strengths of both convolutional and recurrent neural networks to effectively capture spatial and temporal features within electrocardiography signals. To outperform the existent deep learning solutions, we incorporate a systematic data balancing strategy to address the class imbalance often present in the available datasets. Furthermore, we employ an automatic hyperparameter optimization technique to fine-tune the model's parameters for optimal performance. The proposed deep learning architecture, trained on a robust dataset of electrocardiograms, demonstrates exceptional accuracy in classifying multiple arrhythmia types. Our results, with an overall accuracy of 99.60%, surpass previous state-of-the-art methods, highlighting the potential of our approach to improve the efficiency and accuracy of arrhythmia diagnosis in clinical settings.
The high-performance computing (HPC) community has recently seen a substantial diversification of hardware platforms and their associated programming models. From traditional multicore processors to highly specialized accelerators, vendors and tool developers back up the relentless progress of those architectures. In the context of scientific programming, it is fundamental to consider performance portability frameworks, i.e., software tools that allow programmers to write code once and run it on different computer architectures without sacrificing performance. We report here on the benefits and challenges of performance portability using a field-line tracing simulation and a particle-in-cell code, two relevant applications in computational plasma physics with applications to magnetically-confined nuclear-fusion energy research. For these applications we report performance results obtained on four HPC platforms with server-class CPUs from Intel (Xeon) and AMD (EPYC), and high-end GPUs from Nvidia and AMD, including the latest Nvidia H100 GPU and the novel AMD Instinct MI300A APU. Our results show that both Kokkos and OpenMP are powerful tools to achieve performance portability and decent “out-of-the-box” performance, even for the very latest hardware platforms. For our applications, Kokkos provided performance portability to the broadest range of hardware architectures from different vendors.
The impact of high performance computing on technological innovation and scientific discovery is preponderant. Computer simulation, big data analysis, and generative artificial intelligence are often used in trailblazing products or groundbreaking findings. However, an appropriate provisioning of supercomputing resources for the entire spectrum of users is a daunting task. What should be the right hardware to satisfy future needs and demands? We set out to answer that question, particularly in academic environments, where funding schemes are based on availability of grants. The resulting machine, in those institutions, is usually a heterogeneous mix of several architectures and configurations. This paper presents a methodology to guide the next supercomputing purchase based on what is already available and what upcoming needs are anticipated. We use a collection of publicly-available benchmarks and applications to profile a machine. We then use a mathematical model, based on the current profile, to navigate the space of future configurations and suggest future investments. We applied our methodology to Kabré, a small but representative, compute cluster in an academic setting
This study evaluates the performance and reliability of earthquake detection using the EQTransformer, a novel deep learning program that is widely used in seismological observatories and research for enhancing earthquake catalogs. We test the EQTransformer capabilities and uncertainties using seismic data from the Volcanological and Seismological Observatory of Costa Rica and compare two detection options: the simplified method (MseedPredictor) and the complex method (Predictor), the latter incorporating Monte Carlo Dropout, to assess their reproducibility and uncertainty in identifying seismic events. Our analysis focuses on 24 h-duration data that began on February 18, 2023, following a magnitude 5.5 mainshock. Notably, we observed that sequential experiments with identical data and parametrization yield different detections and a varying number of events as a function of time. The results demonstrate that the complex method, which leverages iterative dropout, consistently yields more reproducible and reliable detections than the simplified method, which shows greater variability and is more prone to false positives. This study highlights the critical importance of method selection in deep learning models for seismic event detection, emphasizing the need for rigorous evaluation of detection algorithms to ensure accurate and consistent earthquake catalogs and interpretations. Our findings provide valuable insights for the application of AI tools in seismology, particularly in enhancing the precision and reliability of seismic monitoring efforts.
Plasmids are key determinants in microbial ecology and evolution, facilitating the dissemination of adaptive traits and antibiotic resistance genes (ARGs). Although the molecular mechanisms governing plasmid replication, maintenance, and transfer have been extensively studied, the specific impacts of urbanization-induced pollution on plasmid ecology, diversity, and associated ARGs in tropical regions remain underexplored. This study investigates these dynamics in a tropical aquatic ecosystem, providing novel insights into how pollution shapes plasmid composition and function. In contrast to the observed decrease in chromosomal diversity, we demonstrate that pollution associated with urbanization increases the diversity and taxonomic composition of plasmids within a bacterial community (plasmidome). We analyzed eighteen water and sediment metagenomes, capturing a gradient of pollution and ARG contamination along a tropical urban river. Plasmid and chromosomal diversity profiles were found to be anti-correlated. Plasmid species enrichment along the pollution gradient led to significant compositional differences in water samples, where differentially abundant species suggest plasmid maintenance within specific taxonomic classes. Additionally, the diversity and abundance of ARGs related to the plasmidome increased concomitantly with the intensity of fecal and chemical pollution. These findings highlight the critical need for targeted plasmidome studies to better understand plasmids' environmental spread, as their dynamics are independent of chromosomal patterns. This research is crucial for understanding the consequences of bacterial evolution, particularly in the context of environmental and public health.
Count models are used when the response variable represents the number of specific successes. Depending on the application area, this count may have an excess of zeros, which requires a model that addresses this detail. Zero-inflated models are ideal for such situations. This study analyzes data related to Chagas disease to address both statistical and biological questions. To achieve this, a simulation study is conducted to determine whether variations in sample size affect estimates and if the chosen sampling method influences the performance of metrics used to calculate the probability of cell infection. It is found that balancing the sample size across concentrations using stratified sampling yields the best metric performance values. Additionally, it is concluded that increasing the concentration dose helps to reduce the number of parasites in the cells.
Scientific computing in the exascale era demands increased computational power to solve complex problems across various domains. With the rise of heterogeneous computing architectures the need for vendor-agnostic, performance portability frameworks has been highlighted. Libraries like Kokkos have become essential for enabling high-performance computing applications to execute efficiently across different hardware platforms with minimal code changes. In this direction, this paper presents preliminary time-to-solution results for two representative scientific computing applications: an N-body simulation and a structured grid simulation. Both applications used a distributed memory approach and hardware acceleration through four performance portability frameworks: Kokkos, OpenMP, RAJA, and OCCA. Experiments conducted on a single node of the Polaris supercomputer using four NVIDIA A100 GPUs revealed significant performance variability among frameworks. OCCA demonstrated faster execution times for small-scale validation problems, likely due to JIT compilation, however its lack of optimized reduction algorithms may limit scalability for larger simulations while using its out of the box API. OpenMP performed poorly in the structured grid simulation most likely due to inefficiencies in inter-node data synchronization and communication. These findings highlight the need for further optimization to maximize each framework’s capabilities. Future work will focus on enhancing reduction algorithms, data communication, memory management, as wells as performing scalability studies, and a comprehensive statistical analysis to evaluate and compare framework performance.
Understanding symbiosis between insects and their gut microbiota is critical for uncovering ecological and evolutionary processes. Taxonomic classification of microbial communities using the 16S rRNA gene is an initial standard approach, yet the accuracy and limitations of bioinformatic tools for this purpose, particularly in non-model organisms like beetles, are not well-studied. This study provides a novel benchmarking of three widely used classification tools-QIIME2, Mothur, and VSEARCH-on beetle gut metagenomic data, revealing new insights into their performance. Our results show that all tools achieved an ASV retrieval accuracy above 99 % with consistent taxonomic assignments. Based on these findings, all three tools are suitable for analyzing 16S data from insect gut samples, with the choice depending on specific research objectives or computational limitations. We emphasize the need for careful selection of reference databases to maximize taxonomic accuracy, contributing valuable guidelines for future studies involving complex microbial communities in underexplored hosts.
The Large Hadron Collider (LHC) located at the European Organization for Nuclear Research (CERN) is the largest accelerator in the world, where proton-proton collisions take place at the as-of-today highest energy ever reached. The Large Hadron Collider beauty (LHCb) experiment is located at one of the LHC interaction points, and its detector and readout electronics are currently able to cope with the 40 MHz bunch crossing rate, corresponding to 4 TB/s. A software-only trigger, with the first stage (High-Level Trigger 1) completely running on Graphics Processing Units (GPU) cards, reduces the data rate to 10 GB/s performing partial detector reconstruction and selection. Advanced computing techniques are being used and developed to accelerate information analysis. In this paper, we discuss, and show the first results of a GPU-based sequence for the reconstruction of the LHCb Ring Imaging Cherenkov detectors (RICH) that is currently being implemented. An increase in throughput of up to a factor of 172 with respect to the implementation on CPUs and the achieved 81% performance portability show the potential of using this technology for the rest of the detector reconstruction.
Plasmids play a crucial role in facilitating genetic exchange and enhancing the adaptability of microbial communities. Despite their importance, environmental plasmids remain understudied, particularly those in fragile and underexplored ecosystems such as the deep-sea. In this paper we implemented a bioinformatics pipeline to study the composition, diversity, and functional attributes of plasmid communities (plasmidome) in 81 deep-sea metagenomes from the Tara and Malaspina expeditions, sampled from the Pacific, Atlantic, and Indian Oceans at depths ranging from 270 to 4005 m. We observed an association between depth and plasmid traits, with the 270–1000 m range (mesopelagic samples) exhibiting the highest number of plasmids and the largest plasmid sizes. Plasmids of Alphaproteobacteria and Gammaproteobacteria were predominant across the oceans, particularly in this depth range, which also showed the highest species diversity and abundance of metabolic pathways, including aromatic compound degradation. Surprisingly, relatively few antibiotic resistance genes were found in the deep-sea ecosystem, with most being found in the mesopelagic layer. These included classes such as beta-lactamase, biocide resistance, and aminoglycosides. Our study also identified the MOBP and MOBQ relaxase families as prevalent across various taxonomic classes. This research underscores the importance of studying the plasmidome independently from the chromosomal context. Our limited understanding of the deep-sea’s microbial ecology, especially its plasmidome, necessitates caution in human activities like mining. Such activities could have unforeseen impacts on this largely unexplored ecosystem.