Workload Trends Research Articles

HMM-CPM: a cloud instance resource prediction method tracing the workload trends via hidden Markov model

HMM-CPM: a cloud instance resource prediction method tracing the workload trends via hidden Markov model

Analyzing workload trends for boosting triple stores performance

The Resource Description Framework (RDF) is widely used to model web data. The scale and complexity of the modeled data emphasized performance challenges on the RDF-triple stores. Workload adaption is one important strategy to deal with those challenges on the storage level. Current workload-adaption approaches lack the necessary generalization of the problem and only optimize part of the storage layer with the workload (mostly the replication). This creates a big performance gap within other data structures (e.g. indexes and cache) that could heavily benefit from the same workload adaption strategy. Moreover, the workload statistics are built collectively in most of the current approaches. Thus, the analysis process is unaware of whether workloads’ items are old or recent. However, that does not simulate the temporal trends that exist naturally in user queries which causes the analysis process to lag behind the rapid workload development. We present a novel universal adaption approach to the storage management of a distributed RDF store. The system aims to find optimal data assignments to the different indexes, replications, and join cache within the limited storage space. We present a cost model based on the workload that often contains frequent patterns. The workload is dynamically and continuously analyzed to evaluate predefined rules considering the benefits and costs of all options of assigning data to the storage structures. The objective is to reduce query execution time by letting different data containers compete on the limited storage space. By modeling the workload statistics as time series, we can apply well-known smoothing techniques allowing the importance of the workload to decay over time. That allows the universal adaption to stay tuned with potential changes in the workload trends.

Trends in clinical workload in UK primary care 2005-2019: a retrospective cohort study.

Substantial increases in UK consulting rates, mean consultation duration, and clinical workload were observed between 2007 and 2014. To the authors' knowledge, no analysis of more recent trends in clinical workload has been published to date. This study updates and builds on previous research, identifying underlying changes in population morbidity levels affecting demand for primary health care. To describe the changes in clinical workload in UK primary care since 2005. Retrospective cohort study using GP primary care electronic health records data from 824 UK general practices. Over 500 million anonymised electronic health records were obtained from IQVIA Medical Research Data to examine consulting rates with GPs and practice nurses together with the duration of these consultations to determine total patient-level workload per person-year. Age-standardised mean GP direct (face-to-face and telephone) consulting rates fell steadily by 2.0% a year from 2014 to 2019. Between 2005 and 2019 mean GP direct consulting rates fell by 5.8% overall whereas mean workload per person-year increased by 25.8%, owing in part to a 36.9% increase in mean consultation duration. Indirect GP workload almost tripled over the 15 years, contributing to a 48.3% increase in overall clinical workload per person-year. The proportion of the study population with ≥3 serious chronic conditions increased from 9.7% to 16.1%, accounting for over a third of total clinical workload in 2019. Findings show sustained increases in consulting rates, consultation duration, and clinical workload until 2014. From 2015, however, rising demand for health care and a larger administrative workload have led to capacity constraints as the system nears saturation.

More Tethered to the EHR: EHR Workload Trends Among Academic Primary Care Physicians, 2019-2023.

The purpose of this study is to evaluate recent trends in primary care physician (PCP) electronic health record (EHR) workload. This longitudinal study observed the EHR use of 141 academic PCPs over 4 years (May 2019 to March 2023). Ambulatory full-time equivalency (aFTE), visit volume, and panel size were evaluated. Electronic health record time and inbox message volume were measured per 8 hours of scheduled clinic appointments. From the pre-COVID-19 pandemic year (May 2019 to February 2020) to the most recent study year (April 2022 to March 2023), the average time PCPs spent in the EHR per 8 hours of scheduled clinic appointments increased (+28.4 minutes, 7.8%), as did time in orders (+23.1 minutes, 58.9%), inbox (+14.0 minutes, 24.4%), chart review (+7.2 minutes, 13.0%), notes (+2.9 minutes, 2.3%), outside scheduled hours on days with scheduled appointments (+6.4 minutes, 8.2%), and on unscheduled days (+13.6 minutes, 19.9%). Primary care physicians received more patient medical advice requests (+5.4 messages, 55.5%) and prescription messages (+2.3, 19.5%) per 8 hours of scheduled clinic appointments, but fewer patient calls (-2.8, -10.5%) and results messages (-0.3, -2.7%). While total time in the EHR continued to increase in the final study year (+7.7 minutes, 2.0%), inbox time decreased slightly from the year prior (-2.2 minutes, -3.0%). Primary care physicians' average aFTE decreased 5.2% from 0.66 to 0.63 over 4 years. Primary care physicians' time in the EHR continues to grow. While PCPs' inbox time may be stabilizing, it is still substantially higher than pre-pandemic levels. It is imperative health systems develop strategies to change the EHR workload trajectory to minimize PCPs' occupational stress and mitigate unnecessary reductions in effective physician workforce resulting from the increased EHR burden.

Supply and Demand for Radiation Oncologists in Canada: Workforce Planning Projections From 2020 to 2040

The number of Canadians diagnosed with cancer, and subsequent demand for radiation therapy, are expected to increase over time. This study aimed to update our needs-based workforce planning model to ensure appropriate staffing levels in the future. The supply of radiation oncologists, by age group, sex, and full-time equivalent status, was projected from 2020 to 2040 using a recursive-aging, input-output model developed with seeding parameters derived from national sources. The demand for radiation oncologists until 2040 was estimated using referral patterns for radiation therapy and consultation workload metrics applied to projected annual cancer incident cases to calculate required full-time equivalent positions. Baseline model parameters were also applied to the 2005-2019 workforce and incident case data to evaluate preprojection supply and demand trends. Preprojection trends for 2005 to 2019 revealed accelerated staffing growth that transitioned from a workforce shortage to a surplus state in 2014 followed by substantial growth slowdown in 2016. The model predicts a transient surplus of radiation oncologists until 2026 followed by a projected deficit in subsequent years. Sensitivity analyses using the plausible range for each parameter continued to favor an undersupply, suggesting a trainee shortage unable to meet workforce expansion needs. Considering possible future declining trends in radiotherapy utilization and workload, calculations to inform corrective efforts in resident numbers resulted in 25 entry positions per year, up from 21 per year currently. Geographic distribution of trainees, relative to workforce and cancer incidence distributions, could be improved with more residency positions in Canadian regions outside Ontario. Demand for radiation therapy and radiation oncologists in Canada are expected to grow more quickly than future expansion in staffing levels. Our workforce planning model provides evidence for more trainee requirements to inform stakeholders of possible corrective actions to training programs and recruitment. Further research is needed to explore additional strategies to expand capacity and high-quality delivery of radiation therapy to meet the foreseeable increase in Canadian patients with cancer.

VTGAN: hybrid generative adversarial networks for cloud workload prediction

Efficient resource management approaches have become a fundamental challenge for distributed systems, especially dynamic environment systems such as cloud computing data centers. These approaches aim at load-balancing or minimizing power consumption. Due to the highly dynamic nature of cloud workloads, traditional time series and machine learning models fail to achieve accurate predictions. In this paper, we propose novel hybrid VTGAN models. Our proposed models not only aim at predicting future workloads but also predicting the workload trend (i.e., the upward or downward direction of the workload). Trend classification could be less complex during the decision-making process in resource management approaches. Also, we study the effect of changing the sliding window size and the number of prediction steps. In addition, we investigate the impact of enhancing the features used for training using the technical indicators, Fourier transforms, and wavelet transforms. We validate our models using a real cloud workload dataset. Our results show that VTGAN models outperform traditional deep learning and hybrid models, such as LSTM/GRU and CNN-LSTM/GRU, concerning cloud workload prediction and trend classification. Our proposed model records an upward prediction accuracy ranging from 95.4% to 96.6%.

Multi agent deep reinforcement learning for resource allocation in container‐based clouds environments

AbstractVirtualization enables the deployment of several virtual servers on the same physical layer, critical component of the cloud. As cloud services advance, more apps that use repositories are developed, which adds to the overburden. Containers have evolved into the most reliable and lightweight virtualization technology for cloud services thanks to their flexible sorting, mobility, and scalability. In container‐based clouds, containers can potentially cut data centre energy usage more than virtual machines (VMs) do. Containers are less energy intensive than VMs. Resource allocation is the most prevalent method in cloud systems. However, resource allocation in container‐based clouds (RAC) is innovative and complicated due to its two‐level architecture. This includes the pairing of virtual machines and physical computers with containers. In cloud container services, planner components are essential. This lowers expenses while improving the performance and variety of workloads using cloud resources. The cloud infrastructure resource allocation framework is gaining popularity since it is energy‐efficient and focuses on cloud data management to maximize income and minimize costs. In this paper, we proposed a deep learning‐based architecture capable of achieving high data centre energy efficiency and preventing Service Level Agreement (SLA) violations from deploying green cloud resources. This research describes a hybrid optimum and multi‐agent deep reinforcement learning (MADRL) technique for dynamic task scheduling (DTS) in a container cloud environment. The MADRL‐DTS model for the RAC problem considers VM overheads, VM types, and an affinity restriction. Then, to address the RAC issue, we develop a DTS hyper‐heuristic technique. MADRL‐RAC may give allocation rules by recognizing workload trends and VM types from previous workload traces. Compared to modern procedures, the results demonstrate a significant reduction in energy consumption. The evaluation for energy‐efficient resource allocation is tested in several virtualized environments to get a high power usage effectiveness and CPU usage.

The Visual Behaviour of the Cyclist: Comparison between Simulated and Real Scenarios

Cyclists are one of the main categories of road users particularly exposed to accident risk. The increasing use of this ecological means of transport requires a specific assessment of cyclist safety in terms of traffic flow and human factors. In this study, a particular visual tracking tool has been used to highlight not only the main critical points of the infrastructure, where a high level of distraction is recorded, but also the various interactions with different road users (pedestrians, vehicles, buses, wheelchairs, cyclists). To confirm the critical aspects of the infrastructure and the trend of workload, a similar circuit was reproduced in a bicycle simulator, which also allowed a meaningful comparison of cycling behaviour. The innovative component of this paper is a comparison between a real test, held in Stockholm, and a simulator where the same scenario has been represented, in order to highlight the objective differences in behaviour. The cycling performance was also evaluated both from an objective point of view, with the count of frames related to each category of visualization, and from a subjective one, through the questionnaires. The results show the crossing as a critical aspect because only 4/3% fixation is required for both simulated and real tests to confirm the significance of the comparison between the two experiments. The high attention rate, resulting from frame-by-frame analysis, also points to a clear difference in the perception of users, who feel with a low workload.

AI-Optimized Cost-Aware Design Strategies for Resource-Efficient Applications

In the context of modern computing landscapes marked by escalating resource demands and cost considerations, this paper introduces a novel framework that integrates artificial intelligence (AI) for the creation of resource-efficient applications while maintaining a keen awareness of costs. The imperative to strike a harmonious equilibrium between application performance and expenses has never been more pressing, especially with the proliferation of cloud-based services. In response, our approach capitalizes on AI methodologies to dynamically analyze real-time application requisites, workload trends, and the availability of resources. Central to our methodology is the elevation of cost to a principal design determinant. We devise strategies that dynamically apportion resources, opt for suitable service tiers, and make necessary adjustments to application configurations. This duality of optimizing performance while curtailing expenditure underscores the essence of our approach. Rigorous simulations and empirical evaluations underscore the efficacy of our strategies across diverse scenarios, underscoring substantial cost reductions without compromising the quality of applications.

Prediction models for Clustered Virtual Machines in Data Centers

In data center (DC) environments, machine learning algorithms play an important role in resource management to increase efficiency by means of proper predictive monitoring workload trends and adjusting jobs accordingly. In this paper, we propose a system to predict the CPU usage of virtual machines (VMs) of a DC. Our proposal performs clustering of VMs based on their historical information (i.e., time series) by evaluating several traditional ML algorithms using common statistical features of VM time series, which facilitates grouping VMs with similar behaviors and establishing clusters based on these features. Then, training of representative models is performed to finally choose the one with the lowest mean error per cluster. The simulation results show that by performing clustering and training the model with representative time series, it is indeed possible to obtain a low mean error while reducing the local training time per individual VM.

Creating an ATC knowledge graph in support of the artificial situational awareness system

Automation has been recognized as a possible solution for increasing air traffic controller workload trends. This paper presents a methodology for creating an air traffic control knowledge graph, which is used as part of a hybrid artificial intelligence system for air traffic control operations. The system combines machine learning and symbolic reasoning with the purpose of achieving artificial situational awareness in a narrow domain of en-route air traffic control operations. This approach allows the use of user-defined knowledge alongside existing knowledge repositories. The novel knowledge graph development methodology is universal for any area of air traffic management which relies on the aeronautical information exchange models. In this paper we also present the open-source tools which were developed to make this approach possible and system performance evaluations. Future work should address achieving real-time operation and additional task automation, accompanied by appropriate ontology and graph expansion.

Are We Overdoing It? Changes in Diagnostic Imaging Workload during the Years 2010-2020 including the Impact of the SARS-CoV-2 Pandemic.

Since the 1990s, there has been a significant increase in the number of imaging examinations as well as a related increase in the healthcare expenditure and the exposure of the population to X-rays. This study aimed to analyze the workload trends in radiology during the last decade, including the impact of COVID-19 in a single university hospital in Poland and to identify possible solutions to the challenges that radiology could face in the future. We compared the annual amount of computed tomography (CT), radiography (X-ray), and ultrasound (US) examinations performed between the years 2010 and 2020 and analyzed the changes in the number of practicing radiologists in Poland. The mean number of patients treated in our hospital was 60,727 per year. During the last decade, the number of CT and US examinations nearly doubled (from 87.4 to 155.7 and from 52.1 to 86.5 per 1000 patients in 2010 and 2020 respectively), while X-ray examinations decreased from 115.1 to 96.9 per 1000 patients. The SARS-CoV-2 pandemic did not change the workload trends as more chest examinations were performed. AI, which contributed to the COVID-19 diagnosis, could aid radiologists in the future with the growing workload by increasing the efficiency of radiology departments as well as by potentially minimizing the related costs.

An investigation of changing forensic pathology workloads in the coronial jurisdictions of Australia

The workload of forensic pathologists involves performing autopsies for the coroner. Over the period 2006–2016 it was expected to see a change in autopsy workload of forensic pathologists in Australia due to the increasing number of deaths per year. This study investigated if trends in Australian coronial autopsy workload are related to changes in Coroners Acts. Analysis was performed using data from the National Coronial Information System (NCIS), which was compared with the death data from the Australian Bureau of Statistics (ABS) and Coroners Acts between 2006–2016. In every jurisdiction, the number of deaths per year increased during the study period, but not all jurisdictions observed an increase number of autopsies per year. In three of the eight coronial jurisdictions of Australia, changes in autopsy numbers coincided with changes in Coroners Acts. However, changes in autopsy numbers per year were also observed in without modifications to Coroners Acts. The findings suggest that forensic pathology autopsy workload can be influenced by changes in Coroners Acts and by changes relating to operating polices within individual jurisdictions.

Evaluating Methods for Imputing Missing Data from Longitudinal Monitoring of Athlete Workload.

Missing data can influence calculations of accumulated athlete workload. The objectives were to identify the best single imputation methods and examine workload trends using multiple imputation. External (jumps per hour) and internal (rating of perceived exertion; RPE) workload were recorded for 93 (45 females, 48 males) high school basketball players throughout a season. Recorded data were simulated as missing and imputed using ten imputation methods based on the context of the individual, team and session. Both single imputation and machine learning methods were used to impute the simulated missing data. The difference between the imputed data and the actual workload values was computed as root mean squared error (RMSE). A generalized estimating equation determined the effect of imputation method on RMSE. Multiple imputation of the original dataset, with all known and actual missing workload data, was used to examine trends in longitudinal workload data. Following multiple imputation, a Pearson correlation evaluated the longitudinal association between jump count and sRPE over the season. A single imputation method based on the specific context of the session for which data are missing (team mean) was only outperformed by methods that combine information about the session and the individual (machine learning models). There was a significant and strong association between jump count and sRPE in the original data and imputed datasets using multiple imputation. The amount and nature of the missing data should be considered when choosing a method for single imputation of workload data in youth basketball. Multiple imputation using several predictor variables in a regression model can be used for analyses where workload is accumulated across an entire season.

Effect of Workload After ACL Reconstruction on Rerupture Rates in NBA Players.

Background:Rupture of the anterior cruciate ligament (ACL) is a common and potentially career-altering injury sustained by players in the National Basketball Association (NBA). Strategies have been employed by the league to prevent reinjury of players after ACL reconstruction (ACLR), including minute restrictions and rest games; however, it remains unknown whether workload metrics after ACLR influence the risk for reinjury and revision surgery.Purpose:To evaluate whether workload changes after return to play (RTP) from primary ACLR influences the risk of rerupture in NBA players.Study Design:Case-control study; Level of evidence, 3.Methods:We identified NBA players from 1975 to 2018 who underwent primary ACLR as well as those who required revision ACLR. Primary outcomes included workload measures such as games played, games started, and minutes per game. Secondary outcomes included in-game performance statistics. Statistical analysis was used to compare relative workload and performance 3 years before and 3 years after undergoing primary ACLR. Workload was also compared between the control group of NBA players who underwent primary ACLR and those who required revision ACLR.Results:A total of 68 players who underwent primary ACLR were included, 8 of whom subsequently required revision ACLR. In their first season upon RTP, control players (primary ACLR) demonstrated a significant reduction in all workload metrics relative to the season before injury (P < .001), while the revision group demonstrated an unchanged to increased workload. In a comparison between the primary and revision groups during the first season after RTP, the primary group demonstrated significantly fewer games started (mean ± SD, 22.2 ± 3.0 vs 35.8 ± 8.3; P = .039) and minutes per game (20.5 ± 1.1 vs 27.0 ± 3.1; P = .048) than revision players. The primary ACLR group demonstrated reduced cumulative workload trends for the first 3 years after RTP relative to 3 years before injury, which was not demonstrated in the revision ACLR group, albeit statistically insignificant.Conclusion:Our study found that after ACLR, a reduction in workload parameters relative to preinjury baseline was associated with players who did not sustain rerupture. Further study is required to determine if workload measures following RTP from primary ACLR should be individualized relative to preinjury baseline.

Association Between Physicians' Workload and Prescribing Quality in One Tertiary Hospital in China.

Alarming increasing trends in physician workload have attracted much attention in recent years. Heavy workload may compromise the quality of medication use. Previous studies have identified a series of factors contributing to inappropriate prescribing; however, there is no demonstrated evidence supporting an association between workload and the appropriateness of physicians' prescriptions in China. This study aimed to investigate the relationship between physician workload and prescription quality in a tertiary hospital in Beijing, China. Our study was a single-center, retrospective study, with all outpatient electronic health records extracted from hospital information system of a tertiary hospital in Beijing from July 1 to November 30, 2015. We used outpatient volume in each 5-hour shift as the measure of physician workload. The evaluation of prescribing quality was based on the Rational Drug Use System. Generalized linear models with a γ distribution and a log link were used to explore factors associated with inappropriate prescribing, and we undertook a series of robustness tests with respect to different exclusion criteria. A total of 457,784 prescriptions from 502 physicians were included in the study. Physicians had an average workload of 34.3 (±19.8) patients per shift, and the mean rate of inappropriate prescribing per shift was 14.1% (±14.6%). Higher rates of inappropriate prescribing were associated with heavier workloads (P < 0.001). Physicians who worked in the afternoon, chief physicians, those working in surgical department, males, and those with more than 20-year experience had higher rates of inappropriate prescribing with increasing workload. Heavier workload was associated with higher risk of prescribing inappropriately. It requires great efforts to determine optimal physician workloads and mitigate the potential adverse effects on the prescription quality.

Latency-Based Analytic Approach to Forecast Cloud Workload Trend for Sustainable Datacenters

Cloud datacenters are turning out to be massive energy consumers and environment polluters, which necessitate the need for promoting sustainable computing approaches for achieving environment-friendly datacentre execution. Direct causes of excess energy consumption of the datacentre include running servers at low level of workloads and over-provisioning of server resources to the arriving workloads during execution. To this end, predicting the future workload demands and their respective behaviors at the datacenters are being the focus of recent researches in the context of sustainable datacenters. But prediction analytics of cloud workloads suffer various limitations imposed by the dynamic and unclear characteristics of Cloud workloads. This paper proposes a novel forecasting model named K-means based Rand Variable Learning Rate Backpropagation Neural Network (K-RVLBPNN) for predicting the future workload arrival trend, by exploiting the latency sensitivity characteristics of Cloud workloads, based on a combination of improved K-means clustering algorithm and Backpropagation Neural Network (BPNN) algorithm. Experiments conducted on real-world Cloud datasets shows that the proposed model shows better prediction accuracy, outperforming the traditional Hidden Markov Model, Naive Bayes Classifier, and our earlier RVLBPNN model, respectively.

RTSLPS: Real time server load prediction system for the ever-changing cloud computing environment

Cloud Computing (CC) proposes a multi-tenant framework used by multiple concurrent users, each of which exhibits different and varied behavior. Such heterogeneity shapes a highly fluctuating load and creates new usage patterns overtime at the server level. Virtual Machines (VMs) interference also plays a big part in inducing changes at server load. Server load prediction is deemed crucial to ensure efficient resource usage. The execution of real-time interactive tasks constitutes an important part of CC. Thus, we propose, in this paper, a real-time server load prediction system based on incoming task classification and VM interference detection. The incoming task classification is used to capture the incoming workload trend and VM interference detection aims to capture the interference rate. Finally, load prediction considers server actual resources’ usage, VM interference rate, and incoming workload trend. We propose an improved version of Hoeffding Adaptive Tree (HAT), augmented by ensemble drift detectors. Results show that our Real-Time Server Load Prediction System (RTSLPS) was able to deliver great flexibility dealing with changes and very good accuracy with quick evaluation time and a small memory footprint.

Radiology Education in the 21st Century: Threats and Opportunities

Over the past 25 years, radiology has seen tremendous growth in interpretive demands, including increasing imaging volumes and shorter turnaround times, as well as increased noninterpretive demands often targeting value-adding opportunities. These mounting pressures have led to increased burnout among radiologists nationwide and, in the academic setting, have begun to threaten the core educational mission. Despite these threats, radiology has also proven itself over the years to be a leader when it comes to innovation, and as such, no other field is better suited to confront these challenges with innovative solutions. This article explores the impact these workload trends have had on radiology education and educators, as well as opportunities to confront these challenges.

Taking Stock: The Canadian Association of Radiation Oncology 2017 Radiation Oncologist Workforce Study

To identify and report radiation oncologist (RO) workforce demographics, clinical workload trends, and equipment inventory in Canada. The Canadian Association of Radiation Oncology (CARO) distributed an online survey to RO administrative leaders at 47 Canadian cancer centers providing radiation therapy services from June to December 2017. The survey queried RO staff demographics, clinical workload, and equipment inventory from 2014 to2016. The response rate was 98% and represented 46 of 47 centers for analysis. In 2016, 510 ROs were in practice, with 98 ROs (19.2%) having <1.0 full-time equivalent (FTE) clinical work activities because of administration, research, or part-time employment. Most ROs worked full-time (92.0%), were affiliated with a university (77.5%), and worked in communities with a population >200,000 (84.9%). Approximately half (52.3%) were ≥46 years old. The male-to female ratio was 1.5:1 or higher in all regions of Canada except for Quebec, where there was no gender gap. Part-time employment was more common among female ROs (P < .01). Although FTE staff levels rose steadily between 2014 (456.3) and 2016 (475.8), an increase in patient workload resulted in a rise in the average annual consults per FTE-RO (from 257 to 267). Over a 2-year period, there were 63.5 FTE-recruitments and 44.0 FTE-departures (18.3 FTE-retirements; 25.7 FTE-migration) for a net gain of 19.5 RO-FTEs. An 8.4% increase in FTE staffing to 516 RO-FTEs in 2019 is anticipated, with 22 ROs expected to retire by 2019. There were 251 megavoltage linear accelerators across Canada, with most (39.8%) located in Ontario. Approximately one-fifth (20.7%) of these were older than 10 years and operating beyond the equipment's recommended life span. The Canadian RO workforce demonstrated incremental growth, but risingannual caseloads suggest that radiation therapy demand outpaced RO supplygains. Government funding is required to replace aging equipment in Canada.