Frequent Task Research Articles

Impact of caring for patients with polyhandicap on institutional health care workers’ quality of life: a cross-sectional and longitudinal evaluation

BackgroundProfound intellectual multiple disabilities or polyhandicap (PLH) is defined as a combination of profound mental retardation and serious motor deficits resulting in extreme dependence. Support for these patients is multidisciplinary, complex, and time-consuming. Thus, institutional health care workers (HCWs) face specific working conditions: frequent physical tasks, distressed families, and restricted feedback.ObjectivesWe aimed to identify determinants of quality of life (QoL) of HCWs and to study longitudinal evolution.MethodsThe study used data from the French cohort EVAL-PLH. The participants were institutional HCWs of persons with PLH (age ≥ 3 years at the time of inclusion; age at onset of cerebral lesion <3 years old). Two populations were used: (1) cross sectional study: the sample 1 includes the HCWs assessed at T2 (2020–2021); (2) longitudinal study: the sample 2 includes the HCWs assessed at both T1 (2015–2016) and T2 (2020–2021). The data collected included: sociodemographics, health status, professional variables, and psycho-comportemental aspects. QoL was assessed using WHOQOL-BREF which provides 4 scores.ResultsIn comparison with French norms, the physical and social scores of QoL were significantly lower while the psychological score was significantly higher for (i) the 223 HCWs (participation rate 62%) assessed at T2 and (ii) the 61 HCWs assessed at T1 and T2. The main factors modulating QoL were age, marital status, self-perceived financial difficulties, personal chronic disease, anxiety-mood disorders, nature of coping strategies, and burnout.ConclusionThis study confirms the mixed (negative and positive) impact of caring persons with PLH on the institutional HCWs’ QOL. Main determinants of the HCW’s QOL were: older age, single status, perceived financial difficulties, altered health status, burn out and coping strategies.Clinical trial registration number: NCT02400528.

Analysis of metadynamics simulations by metadynminer.py.

Molecular dynamics simulation is very useful but computationally demanding method of studying dynamics of biomolecular systems. Many enhanced sampling methods were developed in order to obtain the desired results in available computational time. Metadynamics and its variants are common enhanced sampling methods used for this purpose. Metadynamics simulations allow the user to gather large amounts of data, which have to be analyzed to elucidate the properties of the studied system. Here, we present metadynminer.py, a Python package that allows easy and user-friendly analysis and visualization of the results obtained from metadynamics simulations. The built-in functions automate frequent tasks and make the package easy to use for new users, while its many customization options and object-oriented nature allow for integration into specialized data analysis workflows by more advanced users. The 'metadynminer.py' Python package is available under the GPL-3.0 license via PyPi and Conda. The development version is available on GitHub along with issue support (https://github.com/Jan8be/metadynminer.py). Documentation, tutorial and Jupyter Notebook (provided through the public mybinder.org service) are available at https://metadynreporter.cz.

“Which Proof for Which Class?” Factors Influencing the Selection of Mathematical Proofs for Teaching

Selecting a proof for teaching is a frequent task for teachers. However, it is so far unclear, which factors are considered by teachers when selecting proofs. Is the selection based on task and proof characteristics such as the didactical type of proof? Or based on class characteristics such as students’ algebraic skills? Or do teachers’ characteristics such as their proof skills govern their decision? Or is the selection too non-generic for these characteristics to show a meaningful impact?To address these questions, a quasi-experimental study with N = 183 pre-service teachers was conducted to evaluate the influence of each of these factors on their selection of proofs for teaching. Results highlight several significant effects of the abovementioned characteristics and underline that—even at the pre-service level—the selection of proofs is more nuanced than often assumed in prior research and that teachers deliberately and adaptively select proofs for their teaching based on these factors.

Use of SNOMED CT in Large Language Models: Scoping Review.

Large language models (LLMs) have substantially advanced natural language processing (NLP) capabilities but often struggle with knowledge-driven tasks in specialized domains such as biomedicine. Integrating biomedical knowledge sources such as SNOMED CT into LLMs may enhance their performance on biomedical tasks. However, the methodologies and effectiveness of incorporating SNOMED CT into LLMs have not been systematically reviewed. This scoping review aims to examine how SNOMED CT is integrated into LLMs, focusing on (1) the types and components of LLMs being integrated with SNOMED CT, (2) which contents of SNOMED CT are being integrated, and (3) whether this integration improves LLM performance on NLP tasks. Following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines, we searched ACM Digital Library, ACL Anthology, IEEE Xplore, PubMed, and Embase for relevant studies published from 2018 to 2023. Studies were included if they incorporated SNOMED CT into LLM pipelines for natural language understanding or generation tasks. Data on LLM types, SNOMED CT integration methods, end tasks, and performance metrics were extracted and synthesized. The review included 37 studies. Bidirectional Encoder Representations from Transformers and its biomedical variants were the most commonly used LLMs. Three main approaches for integrating SNOMED CT were identified: (1) incorporating SNOMED CT into LLM inputs (28/37, 76%), primarily using concept descriptions to expand training corpora; (2) integrating SNOMED CT into additional fusion modules (5/37, 14%); and (3) using SNOMED CT as an external knowledge retriever during inference (5/37, 14%). The most frequent end task was medical concept normalization (15/37, 41%), followed by entity extraction or typing and classification. While most studies (17/19, 89%) reported performance improvements after SNOMED CT integration, only a small fraction (19/37, 51%) provided direct comparisons. The reported gains varied widely across different metrics and tasks, ranging from 0.87% to 131.66%. However, some studies showed either no improvement or a decline in certain performance metrics. This review demonstrates diverse approaches for integrating SNOMED CT into LLMs, with a focus on using concept descriptions to enhance biomedical language understanding and generation. While the results suggest potential benefits of SNOMED CT integration, the lack of standardized evaluation methods and comprehensive performance reporting hinders definitive conclusions about its effectiveness. Future research should prioritize consistent reporting of performance comparisons and explore more sophisticated methods for incorporating SNOMED CT's relational structure into LLMs. In addition, the biomedical NLP community should develop standardized evaluation frameworks to better assess the impact of ontology integration on LLM performance.

Presenting the framework of the whole slide image file Babel fish: An OCR-based file labeling tool

IntroductionMetadata extraction from digitized slides or whole slide image files is a frequent, laborious, and tedious task. In this work, we present a tool to automatically extract all relevant slide information, such as case number, year, slide number, block number, and staining from the macro-images of the scanned slide.We named the tool Babel fish as it helps translate relevant information printed on the slide. It is written to contain certain basic assumptions regarding, for example, the location of certain information. This can be adapted to the respective location. The extracted metadata can then be used to sort digital slides into databases or to link them with associated case IDs from laboratory information systems. Material and methodsThe tool is based on optical character recognition (OCR). For most information, the easyOCR tool is used. For the block number and cases with insufficient results in the first OCR round, a second OCR with pytesseract is applied.Two datasets are used: one for tool development has 342 slides; and another for one for testing has 110 slides. ResultsFor the testing set, the overall accuracy for retrieving all relevant information per slide is 0.982. Of note, the accuracy for most information parts is 1.000, whereas the accuracy for the block number detection is 0.982. ConclusionThe Babel fish tool can be used to rename vast amounts of whole slide image files in an image analysis pipeline. Furthermore, it could be an essential part of DICOM conversion pipelines, as it extracts relevant metadata like case number, year, block ID, and staining.

Fg-ORKA: fast and gridless reconstruction of moving and deforming objects in multidimensional data

Abstract Identifying and tracking objects over multiple observations is a frequent task in many applications. Traffic monitoring requires the tracking of vehicles or pedestrians in video data and geophysical exploration relies on identifying seismic wave fronts from data of multiple sensors, only to mention two examples. In many cases, the object changes its shape or position within the given data from one observation to another. Vehicles can change their position and angle relative to the camera while seismic waves have different arrival times, frequencies, or intensities depending on the sensor position. This complicates the task at hand.

In a previous work, the authors presented a new algorithm to solve this problem - Object reconstruction using K-approximation (ORKA). This algorithm is hindered by two conflicting limitations: the tracked movement is limited by the sampling grid while the complexity increases exponentially with the resolution. We introduce an iterative variant of the ORKA algorithm that is able to overcome this conflict. We also give a brief introduction on the original ORKA algorithm. Knowledge of the previous work is thus not required.

We give theoretical error bounds and a complexity analysis which we validate with several numerical experiments. Moreover, we discuss the influence of different parameter choices in detail. The results clearly show that the iterative approach can outperform ORKA in both accuracy and efficiency. On the example of video processing we show that the new method can be applied where the original algorithm is too time and memory intensive. Furthermore, we demonstrate on seismic exploration data that we are now able to recover much finer details on the wave front movement then before.

Aircraft Skin Machine Learning-Based Defect Detection and Size Estimation in Visual Inspections

Aircraft maintenance is a complex process that requires a highly trained, qualified, and experienced team. The most frequent task in this process is the visual inspection of the airframe structure and engine for surface and sub-surface cracks, impact damage, corrosion, and other irregularities. Automated defect detection is a valuable tool for maintenance engineers to ensure safety and condition monitoring. The proposed approach is to process the captured feedback using various deep learning architectures to achieve the highest performance defect detections. Additionally, an algorithm is proposed to estimate the size of the detected defect. The team collaborated with TUI’s Airline Maintenance Team at Luton Airport, allowing us to fly a drone inside the hangar and use handheld cameras to collect representative data from their aircraft fleet. After a comprehensive dataset was constructed, multiple deep-learning architectures were developed and evaluated. The models were optimized for detecting various aircraft skin defects, with a focus on the challenging task of dent detection. The size estimation approach was evaluated in both controlled laboratory conditions and real-world hangar environments, providing insights into practical implementation challenges.

Comprehensive survey of artificial intelligence techniques and strategies for climate change mitigation

With the galloping progress of the changing climates all around the world, Machine Learning (ML) approaches have been prevalently studied in many types of research in this area. ML is a robust tool for acquiring perspectives from data. In this paper, we elaborate on climate change mitigation issues and ML approaches leveraged to solve these issues and aid in the improvement and function of sustainable energy systems. ML has been employed in multiple applications and many scopes of climate subjects such as ecosystems, agriculture, buildings and cities, industry, and transportation. So, a Systematic Literature Review (SLR) is applied to explore and evaluate findings from related research. In this paper, we propose a novel taxonomy of Deep Learning (DL) method applications for climate change mitigation, a comprehensive analysis that has not been conducted before. We evaluated these methods based on critical parameters such as accuracy, scalability, and interpretability and quantitatively compared their results. This analysis provides new insights into the effectiveness and reliability of DL methods in addressing climate change challenges. We classified climate change ML methods into six key customizable groups: ecosystems, industry, buildings and cities, transportation, agriculture, and hybrid applications. Afterward, state-of-the-art research on ML mechanisms and applications for climate change mitigation issues has been highlighted. In addition, many problems and issues related to ML implementation for climate change have been mapped, which are predicted to stimulate more researchers to manage the future disastrous effects of climate change. Based on the findings, most of the papers utilized Python as the most common simulation environment 38.5 % of the time. In addition, most of the methods were analyzed and evaluated in terms of some parameters, namely accuracy, latency, adaptability, and scalability, respectively. Lastly, classification is the most frequent ML task within climate change mitigation, accounting for 40 % of the total. Furthermore, Convolutional Neural Networks (CNNs) are the most widely utilized approach for a variety of applications.

Contributing Factors to Central Line-associated Bloodstream Infections and Catheter-associated Urinary Tract Infections

Background: Central line-associated bloodstream infections (CLABSI) and catheter-associated urinary tract infections (CAUTI) are key healthcare-associated infection (HAI) quality metrics. In this qualitative analysis, we aimed to identify common issues contributing to CLABSIs and CAUTIs occurring during the COVID-19 pandemic. Methods: In an academic healthcare network in Atlanta, GA, four hospitals perform real-time, apparent cause analyses (ACAs) for all CLABSIs and CAUTIs. Contributing factors are entered as free text into an electronic database. We analyzed data from 8/2020–8/2022. We first performed a qualitative open card sort of all reported contributing factors to CLABSI and created a novel framework based on mutually defined critical tasks (e.g., line insertion) and cross-cutting issues (e.g., communication breakdown). Contributing factors could describe ≥1 critical task and/or ≥1 cross-cutting issue. After establishing interrater reliability, a multidisciplinary group applied this framework to classify each contributing factor. For CAUTI, we used the same set of cross-cutting issues but identified new critical tasks via open card sorting. We then used the framework to classify each CAUTI contributing factor. We used descriptive statistics to identify frequent critical tasks and cross-cutting issues. Results: We reviewed 350 CLABSI ACAs with 602 contributing factors and 240 CAUTI ACAs with 405 contributing factors (Figure 1). Our classification framework comprised 11 cross-cutting issues and 9 critical tasks for CLABSI and 7 critical tasks for CAUTI (Figure 2). CLABSI: The critical tasks most often reported were bathing (19%), central line dressing maintenance (15%), and assessing central line indication (8%; Figure 3). Within these tasks, the most frequent issues described for bathing were the task not being performed (20%) and unclear documentation (18%); for dressing maintenance, the task was not performed (15%), not documented (15%), or poorly performed due to lack of competency (15%); and for assessing line indication, there was frequent communication breakdown (33%). CAUTI: The critical tasks most often reported were urinary catheter care (26%) and assessing the indication for urinary catheter (22%; Figure 4). Within these tasks, urinary catheter care was frequently not documented (38%) or not performed (16%); assessing urinary catheter necessity was often not documented (29%) or involved breakdown of communication (19%). Conclusion: We created a novel framework to evaluate common causes of HAIs in an academic healthcare network. This framework can be used to identify and track gaps over time and to develop quality improvement initiatives targeting key tasks and associated factors, such as communication difficulties when assessing device indications.Disclosure: Colleen Kraft: Consultant - REbiotix/Ferring; Scientific Advisory Board - Seres, LLC

Slope multi-step excavation displacement prediction surrogate model based on a long short-term memory neural network: for small sample data and multi-feature multi-task learning

ABSTRACT Machine learning-based surrogate models have become the preferred approach for large-scale and frequent simulation tasks due to its significant improvement in computational efficiency. In order to overcome the potential effects of learning with small sample data and the challenges of multi-feature multi-task learning, we developed a novel deep learning long short-term memory (LSTM) model. Taking slope excavation displacement prediction as a case study, we employed the Latin hypercube sampling method to generate a synthetic dataset for training LSTM and other mainstream models. Experimental results demonstrate that the model's prediction accuracy decreases with a reduction in sample size, while support vector regression (SVR), back propagation neural network (BPNN), LSTM and Gaussian process regression (GPR) demonstrate a stronger resistance. It is feasible to utilise excavation features as model inputs to establish a unified multi-step excavation model, but the accuracy of the SVR model decreased by 32.5% after supplementing excavation features. Even when the sample size is less than 50, both LSTM and GPR exhibit excellent performance, achieving model R-squared and RMSE surpassing 0.99 and 0.07 mm. However, when addressing multi-output learning tasks, LSTM stands out as the optimal choice. This study will assist researchers or engineers in swiftly selecting appropriate surrogate models.

A Review Paper on Paint Spraying Robot

Abstract: Painting interior walls is a frequent construction task that takes a lot of time and work. Robotic painting was introduced to replace human manual activity, improving accuracy, efficiency, and lowering costs. In this study, We present an independent robot that paints walls. robot that uses a cascade lift mechanism to enable it to use a paint sprayer to paint a room's interior walls. The paint sprayer can reach the necessary heights With the assistance of of this mechanism for cascading lifts mechanism. With two degrees of freedom (DOF), The robot moves fluidly in each of the six directions. thanks to the DC powered mecanum wheels mounted down to its foundation. Ultrasonic sensors are used by the robot to measure distance, make adjustments to the walls, and determine if the mister has reached the wall's summit. The robot's mecanum wheels, ultrasonic sensors, and other components are all managed by the master controller. The AC power supply powers the entire system Color is sprayed on to manage the amount of color; sprayers are employed for this purpose. Sprayers need to split liquid into droplets that are the right size, distribute them evenly across the surface, and manage the volume of liquid to prevent overapplication. One major issue thataffects workers everywhere is disease control

Few-shot cow identification via meta-learning

Cow identification is a prerequisite for precision livestock farming. Biometric-based methods have made significant progress in cow identification. However, substantial labelling costs and frequent identification task changes are still hamper model application. In this work, a novel method called “MFCI” was proposed to achieve accurate cow identification under few-shot and task-changing conditions. Specifically, the proposed method comprises two components: cow location and cow identification. First, an improved YOLOv5n with Ghost module was adopted to quickly detect cow locations in images. Then, the Model-Agnostic Meta-Learning (MAML) framework was introduced for accurate identification under few-shot conditions and for fast adaptation to frequent changes in individual cows. Moreover, an autoencoder was adopted to allow Base-Learner learn more generalized features by combining both supervised and unsupervised approaches. The experimental results showed that the proposed cow location model achieved a mAP of 99.5 %. The proposed cow identification model attained an accuracy of 90.43 % with only five samples per cow for 20 cows, outperforming other state-of-the-art methods. The results demonstrate the broad applicability and significant value of the proposed method.

Trust-Aware Reflective Control for Fault-Resilient Dynamic Task Response in Human–Swarm Cooperation

Due to the complexity of real-world deployments, a robot swarm is required to dynamically respond to tasks such as tracking multiple vehicles and continuously searching for victims. Frequent task assignments eliminate the need for system calibration time, but they also introduce uncertainty from previous tasks, which can undermine swarm performance. Therefore, responding to dynamic tasks presents a significant challenge for a robot swarm compared to handling tasks one at a time. In human–human cooperation, trust plays a crucial role in understanding each other’s performance expectations and adjusting one’s behavior for better cooperation. Taking inspiration from human trust, this paper introduces a trust-aware reflective control method called “Trust-R”. Trust-R, based on a weighted mean subsequence reduced algorithm (WMSR) and human trust modeling, enables a swarm to self-reflect on its performance from a human perspective. It proactively corrects faulty behaviors at an early stage before human intervention, mitigating the negative influence of uncertainty accumulated from dynamic tasks. Three typical task scenarios {Scenario 1: flocking to the assigned destination; Scenario 2: a transition between destinations; and Scenario 3: emergent response} were designed in the real-gravity simulation environment, and a human user study with 145 volunteers was conducted. Trust-R significantly improves both swarm performance and trust in dynamic task scenarios, marking a pivotal step forward in integrating trust dynamics into swarm robotics.

Influence of spectral shaping and tube voltage modulation in ultralow-dose computed tomography of the abdomen

PurposeUnenhanced abdominal CT constitutes the diagnostic standard of care in suspected urolithiasis. Aiming to identify potential for radiation dose reduction in this frequent imaging task, this experimental study compares the effect of spectral shaping and tube voltage modulation on image quality.MethodsUsing a third-generation dual-source CT, eight cadaveric specimens were scanned with varying tube voltage settings with and without tin filter application (Sn 150, Sn 100, 120, 100, and 80 kVp) at three dose levels (3 mGy: standard; 1 mGy: low; 0.5 mGy: ultralow). Image quality was assessed quantitatively by calculation of signal-to-noise ratios (SNR) for various tissues (spleen, kidney, trabecular bone, fat) and subjectively by three independent radiologists based on a seven-point rating scale (7 = excellent; 1 = very poor).ResultsIrrespective of dose level, Sn 100 kVp resulted in the highest SNR of all tube voltage settings. In direct comparison to Sn 150 kVp, superior SNR was ascertained for spleen (p ≤ 0.004) and kidney tissue (p ≤ 0.009). In ultralow-dose scans, subjective image quality of Sn 100 kVp (median score 3; interquartile range 3–3) was higher compared with conventional imaging at 120 kVp (2; 2–2), 100 kVp (1; 1–2), and 80 kVp (1; 1–1) (all p < 0.001). Indicated by an intraclass correlation coefficient of 0.945 (95% confidence interval: 0.927–0.960), interrater reliability was excellent.ConclusionsIn abdominal CT with maximised dose reduction, tin prefiltration at 100 kVp allows for superior image quality over Sn 150 kVp and conventional imaging without spectral shaping.

Concurrent expectation and experience-based metacontrol: EEG insights and the role of working memory capacity.

We investigated the simultaneous influence of expectation and experience on metacontrol, which we define as the instantiation of context-specific control states. These states could entail heightened control states in preparation for frequent task switching or lowered control states for task repetition. Specifically, we examined whether "expectations" regarding future control demands prompt proactive metacontrol, while "experiences" with items associated with specific control demands facilitate reactive metacontrol. In Experiment 1, we utilized EEG with a high temporal resolution to differentiate between brain activities associated with proactive and reactive metacontrol. We successfully observed cue-locked and image-locked ERP patterns associated with proactive and reactive metacontrol, respectively, supporting concurrent instantiation of two metacontrol modes. In Experiment 2, we focused on individual differences to investigate the modulatory role of working memory capacity (WMC) in the concurrent instantiation of two metacontrol modes. Our findings revealed that individuals with higher WMC exhibited enhanced proactive metacontrol, indicated by smaller response time variability (RTV). Additionally, individuals with higher WMC showed a lower tendency to rely on reactive metacontrol, indicated by a smaller item-specific switch probability (ISSP) effect. In conclusion, our results suggest that proactive and reactive metacontrol can coexist, but their interplay is influenced by individuals' WMC. Higher WMC promotes the use of proactive metacontrol while attenuating reliance on reactive metacontrol. This study provides insights into the interplay between proactive and reactive metacontrol and highlights the impact of WMC on their concurrent instantiation.

Random pinhole attenuator for high-power laser beams

Abstract The intensity attenuation of a high-power laser is a frequent task in the measurements of optical science. Laser intensity can be attenuated by inserting an optical element, such as a partial reflector, polarizer or absorption filter. These devices are, however, not always easily applicable, especially in the case of ultra-high-power lasers, because they can alter the characteristics of a laser beam or become easily damaged. In this study, we demonstrated that the intensity of a laser beam could be effectively attenuated using a random pinhole attenuator (RPA), a device with randomly distributed pinholes, without changing the beam properties. With this device, a multi-PW laser beam was successfully attenuated and the focused beam profile was measured without any alterations of its characteristics. In addition, it was confirmed that the temporal profile of a laser pulse, including the spectral phase, was preserved. Consequently, the RPA possesses significant potential for a wide range of applications.

PROWIS: A Visual Approach for Building, Managing, and Analyzing Weather Simulation Ensembles at Runtime.

Weather forecasting is essential for decision-making and is usually performed using numerical modeling. Numerical weather models, in turn, are complex tools that require specialized training and laborious setup and are challenging even for weather experts. Moreover, weather simulations are data-intensive computations and may take hours to days to complete. When the simulation is finished, the experts face challenges analyzing its outputs, a large mass of spatiotemporal and multivariate data. From the simulation setup to the analysis of results, working with weather simulations involves several manual and error-prone steps. The complexity of the problem increases exponentially when the experts must deal with ensembles of simulations, a frequent task in their daily duties. To tackle these challenges, we propose ProWis: an interactive and provenance-oriented system to help weather experts build, manage, and analyze simulation ensembles at runtime. Our system follows a human-in-the-loop approach to enable the exploration of multiple atmospheric variables and weather scenarios. ProWis was built in close collaboration with weather experts, and we demonstrate its effectiveness by presenting two case studies of rainfall events in Brazil.

A multi-skill RCPSP variant for persons with disabilities in sheltered workshops

Sheltered workshops face the challenge of meeting their manufacturing objectives while considering the individual competencies and assistance needs of persons with disabilities. Moreover, work processes in sheltered workshops are weakly structured, allowing for frequent task interruptions and changes based on the preferences of impaired work persons. While the Industry 4.0 literature provides many real-time scheduling algorithms for incorporating multiple objectives and constraints, these algorithms fall short of the characteristics of sheltered workshops. The resource-constrained project scheduling problem (RCPSP) is an optimization problem for computing manufacturing plans considering multiple objectives and resource constraints. Among many different RCPSP variants proposed in the literature, the multi-skill RCPSP (MSRCPSP) variant considers the individual skills of work persons when generating manufacturing plans. With the ongoing digital transformation of enterprises, new assistance systems enter the market, providing individual support to impaired work persons and compensating for lacking skills and abilities.This paper proposes an MSRCPSP variant that incorporates assistance systems and learning tasks compensating for competence gaps in the skill matrix of impaired work persons. Furthermore, it decomposes tasks into individual work objects, which accounts for frequent task interruptions and task preferences of work persons. The algorithm is described and demonstrated using a manufacturing data set from an actual sheltered workshop. In a small evaluation study, the algorithm is tested by scheduling two impaired work persons in the assembly department of a medium-sized manufacturing company in the primary labor market. The evaluation study proves its real-world applicability and the suitability of scheduling algorithms for participation and inclusive work.

Applying Natural Language Processing to Textual Data From Clinical Data Warehouses: Systematic Review.

In recent years, health data collected during the clinical care process have been often repurposed for secondary use through clinical data warehouses (CDWs), which interconnect disparate data from different sources. A large amount of information of high clinical value is stored in unstructured text format. Natural language processing (NLP), which implements algorithms that can operate on massive unstructured textual data, has the potential to structure the data and make clinical information more accessible. The aim of this review was to provide an overview of studies applying NLP to textual data from CDWs. It focuses on identifying the (1) NLP tasks applied to data from CDWs and (2) NLP methods used to tackle these tasks. This review was performed according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. We searched for relevant articles in 3 bibliographic databases: PubMed, Google Scholar, and ACL Anthology. We reviewed the titles and abstracts and included articles according to the following inclusion criteria: (1) focus on NLP applied to textual data from CDWs, (2) articles published between 1995 and 2021, and (3) written in English. We identified 1353 articles, of which 194 (14.34%) met the inclusion criteria. Among all identified NLP tasks in the included papers, information extraction from clinical text (112/194, 57.7%) and the identification of patients (51/194, 26.3%) were the most frequent tasks. To address the various tasks, symbolic methods were the most common NLP methods (124/232, 53.4%), showing that some tasks can be partially achieved with classical NLP techniques, such as regular expressions or pattern matching that exploit specialized lexica, such as drug lists and terminologies. Machine learning (70/232, 30.2%) and deep learning (38/232, 16.4%) have been increasingly used in recent years, including the most recent approaches based on transformers. NLP methods were mostly applied to English language data (153/194, 78.9%). CDWs are central to the secondary use of clinical texts for research purposes. Although the use of NLP on data from CDWs is growing, there remain challenges in this field, especially with regard to languages other than English. Clinical NLP is an effective strategy for accessing, extracting, and transforming data from CDWs. Information retrieved with NLP can assist in clinical research and have an impact on clinical practice.

Finding recently persistent flows in high-speed packet streams based on cuckoo filter

In high-speed networks, flow-level traffic measurement is an essential tool to understand how network bandwidth is consumed and support the detection of anomalous traffic. While many prior work focuses on tracking the frequent flows (a.k.a. heavy hitters), in this paper, we put more focus on tracking the persistent flows, which jointly consider the frequency, duration and regularity of the packet arrival events of a flow. Although this more generalized metric called persistence has been defined before, it is still unknown how to use limited memory on data plane to monitor the top-k persistent flows in a recent time interval. In this paper, we propose an algorithm named PFD-DW built upon cuckoo filter (an improved version of hash table with better memory efficiency), to monitor the top-k persistent flows under the time-decaying window. It can be regarded as a variant of cuckoo filter, which transforms each bucket into a bucket-level min-heap. Its advantage is that, when the table is full and a packet of a new flow arrives, it can select the least persistent flow along the cuckoo kicking path as the victim of flow replacement. We deliberately avoid the scanning of the entire table to keep the high time efficiency. Based on real-world network traffic traces, we evaluate the performance of our DAKP-CF, a degraded version of PFD-DW that only considers each flow’s packet arrival frequency. The results show that it outperforms the existing algorithms for the top-k frequent flow identification task. It provides nearly 98% identification accuracy with roughly 50% less memory cost. We also evaluate our PFD-DW algorithm for the more generalized task of identifying the recently persistent flows. It provides 93% identification rate of top-3000 persistent flows using only 576KB memory, while attaining 1.5% persistence estimation error. Its memory cost is reduced by 97% than another proposed solution PFD-SW under sliding time window. We have also developed a prototype of PFD-DW, based on the Fd.io VPP software switch accelerated by Intel DPDK.