Search

Telemedicine is frequently used to provide remote neurological expertise for acute stroke workup and was associated with better functional outcomes when combined with a stroke unit system-of-care. We investigated whether such system-of-care yields additional benefits when implemented on top of neurological competence already available onsite.Quality improvement measures were implemented within a 'hub-and-spoke' teleneurology network in 11 hospitals already provided with onsite or telestroke expertise. Measures included dedicated units for neurological emergencies, standardization of procedures, multiprofessional training, and quality-of-care monitoring. Intervention effects were investigated in a controlled study enrolling patients insured at 3 participating statutory health insurances diagnosed with acute stroke or other neurological emergencies. Outcomes during the intervention period between November 2017 and February 2020 were compared with those pre-intervention between October 2014 and March 2017. To control for temporal trends, we compared outcomes of patients with respective diagnoses in 11 hospitals of the same region. Primary outcome was the composite of up-to-90-day death, new disability with the need of ambulatory or nursing home care, expressed by adjusted hazard ratio (aHR).We included 1,418 patients post-implementation (55% female, mean age 76.7 ± 12.8 year) and 2,306 patients pre-implementation (56%, 75.8 ± 13.0 year, respectively). The primary outcome occurred in 479/1,418 (33.8%) patients post-implementation and in 829/2,306 (35.9%) pre-implementation. The aHR for the primary outcome was 0.89 (95% confidence interval [CI]: 0.79-0.99, p = 0.04) with no improvement seen in non-participating hospitals between post- versus pre-implementation periods (aHR 1.04; 95% CI: 0.95-1.15).Implementation of a multicomponent system-of-care was associated with a lower risk of poor outcomes. ANN NEUROL 2022.

Background: Post-stroke delirium (PSD) is a modifiable predictor for worse outcome in stroke. Knowledge of its risk factors would facilitate clinical management of affected patients, but recently updated national guidelines consider available evidence insufficient. Aims: The study aimed to establish risk factors for PSD incidence and duration using high-frequency screening. Methods: We prospectively investigated patients with ischemic stroke admitted within 24 h. Patients were screened twice daily for the presence of PSD throughout the treatment period. Sociodemographic, treatment-related, and neuroimaging characteristics were evaluated as predictors of either PSD incidence (odds ratios (OR)) or duration (PSD days/unit of the predictor, b), using logistic and linear regression models, respectively. Results: PSD occurred in 55/141 patients (age = 73.8 ± 10.4 years, 61 female, National Institutes of Health Stroke Scale (NIHSS) = 6.4 ± 6.5). Age (odds ratio (OR) = 1.06 (95% confidence interval (CI): 1.02–1.10), b = 0.08 (95% CI = 0.04–0.13)), and male gender ( b = 0.99 (95% CI = 0.05–1.93)) were significant non-modifiable risk factors. In a multivariable model adjusted for age and gender, presence of pain (OR mvar = 1.75 (95% CI = 1.12–2.74)), urinary catheter (OR mvar = 3.16 (95% CI = 1.10–9.14)) and post-stroke infection (PSI; OR mvar = 4.43 (95% CI = 1.09–18.01)) were predictors of PSD incidence. PSD duration was impacted by presence of pain ( b mvar = 0.49 (95% CI = 0.19–0.81)), urinary catheter ( b mvar = 1.03 (95% CI = 0.01–2.07)), intravenous line ( b mvar = 0.36 (95% CI = 0.16–0.57)), and PSI ( b mvar = 1.60 (95% CI = 0.42–2.78)). PSD (OR = 3.53 (95% CI = 1.48–5.57)) and PSI (OR = 5.29 (95% CI = 2.92–7.66)) independently predicted inferior NIHSS at discharge. Insular and basal ganglia lesions increased the PSD risk about four- to eight-fold. Discussion/Conclusion: This study identified modifiable risk factors, the management of which might reduce the negative impact PSD has on outcome.

Background and Purpose: Poststroke delirium (PSD) is an independent predictor of unfavorable outcome. Despite its individual and socioeconomic burden, its frequency, clinical course, and routine detection remain unresolved. This study aimed to assess psychometric properties of established delirium screening tools and investigate the natural course of PSD. Methods: This study investigated patients presenting with high-risk transient ischemic attacks or ischemic stroke within 24 hours during a 3-month period. Twice-daily screenings for PSD were done using the confusion assessment method, nursing delirium scale, and rapid delirium assessment, and evaluated for noninferiority against Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition criteria. We investigated demographic and stroke characteristics as predictors of PSD, neurological deficits as predictors of false screening results, and conducted a simulation study to estimate the best timing to identify PSD. Results: We enrolled 141 patients (73.8±10.4 years of age, 61 female) with a mean National Institutes of Health Stroke Scale score of 6.4±6.5. Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition based PSD incidence was 39%, which manifested within 24 hours in 25% and 72 hours in almost all cases. The confusion assessment method was the only screening tool noninferior to Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition ratings providing a sensitivity of 82% and specificity of 80%. Age (odds ratio, 1.07 [1.02–1.13] per year, P =0.004) and National Institutes of Health Stroke Scale (odds ratio, 1.24 [1.15–1.34] per point, P P P =0.008). Simulations revealed that one in 4 cases is missed with less than daily screenings. Conclusions: PSD is a common complication of stroke and transient ischemic attack. Detection is challenged by confounding effects such as focal neurological deficits and the necessity for at least daily screenings. Future studies are required to investigate implementation of these findings in clinical routine. Registration: URL: https://www.clinicaltrials.gov . Unique identifier: NCT03930719.

A scientific workflow is a set of interdependent compute tasks orchestrating large scale data analyses or in-silico experiments. Workflows often comprise thousands of tasks with heterogeneous resource requirements that need to be executed on distributed resources. Many workflow engines solve parallelization by submitting tasks to a batch scheduling system, which requires resource usage estimates that have to be provided by users. We investigate the possibility to improve upon inaccurate user estimates by incorporating an online feedback loop between workflow scheduling, resource usage prediction, and measurement.Our approach can learn resource usage of arbitrary type; in this paper, we demonstrate its effectiveness by predicting peak memory usage of tasks, as it is an especially sensitive resource type that leads to task termination if underestimated and leads to decreased throughput if overestimated.We compare online versions of standard machine learning models for peak memory usage prediction and analyze their interactions with different workflow scheduling strategies. By means of extensive simulation experiments, we found that the proposed feedback mechanism improves resource utilization and execution times compared to typical user estimates.

List scheduling is an approach to task graph scheduling that has been shown to work well for scheduling tasks with data dependencies on heterogeneous resources. Key to the performance of a list scheduling heuristic is its method to prioritize the tasks, and various ranking schemes have been proposed in the literature. We propose a method that combines multiple random rankings instead of a using a deterministic ranking scheme. We introduce L-Orders, which are a subset of all topological orders of a directed acyclic graph. L-Orders can be used to explore targeted regions of the space of all topological orders. Using the observation that the makespans in one such region are often approximately normal distributed, we estimate the expected time to solution improvement in certain regions of the search space. We combine targeted search and improvement time estimations into a time budgeted search algorithm that balances exploration and exploitation of the search space. In 40,500 experiments, our schedules are 5% shorter on average and up to 40% shorter in extreme cases than schedules produced by HEFT.

In this chapter, we discuss the scientific question whether the clustering of time series based on RQA measures leads to an interpretable clustering structure when analyzed by human experts. We are not aware of studies answering this scientific question. Answering it is the crucial first step in the development of a Visual Analytics approach that support users to explore large sets of time series.