Search

Social learning is common in nature, yet cumulative culture (where knowledge and technology increase in complexity and diversity over time) appears restricted to humans. To understand why, we organized a computer tournament in which programmed entries specified when to learn new knowledge and when to refine (i.e. improve) existing knowledge. The tournament revealed a 'refinement paradox': refined behavior afforded higher payoffs as individuals converged on a small number of successful behavioral variants, but refining did not generally pay. Paradoxically, entries that refined only in certain conditions did best during behavioral improvement, while simple copying entries thrived when refinement levels were high. Cumulative cultural evolution may be rare in part because sophisticated strategies for improving knowledge and technology are initially advantageous, yet complex culture, once achieved, favors conformity, blind imitation and hyper-credulity., Competing Interests: The authors have declared that no competing interests exist., (Copyright: © 2024 Miu et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.)

Purpose: We aimed to validate hyperintense vessel sign (HVS) on FLAIR imaging or posterior cerebral artery (PCA) laterality on MR angiography beyond 4.5 hours after stroke onset., Materials and Methods: Data from acute ischemic stroke patients with internal carotid or middle cerebral artery occlusion who underwent CT perfusion imaging at baseline, follow-up MR perfusion imaging and angiography within 30 hours after stroke, without effective recanalization on follow-up imaging, were analysed retrospectively. Patients were separately classified as high or low HVS (>5 or ≤5 slices of HVS), and PCA laterality positive or negative group. We compared core and penumbra volumes at follow-up imaging and neurological outcomes between high or low HVS group, and between PCA laterality positive or negative group., Results: Of 49 patients analyzed, four patients with artifacts were excluded and 45 were classified into high (n = 23) or low (n = 22) HVS group. High group had a smaller core volume (median 32 ml versus 109 ml, p = 0.004), larger penumbra volume at follow-up (68 ml versus 0 ml, p = 0.001), and better outcomes (modified Rankin Scale at three months, 3 versus 5, p = 0.03). For PCA laterality analysis, 1 patient with previously occluded PCA was excluded and 48 patients were classified as positive (n = 22) or negative (n = 26). Positive group had larger core volume (116 ml versus 37 ml), and no significant differences in penumbral volumes or outcomes., Conclusion: Prominent HVS in later time was associated with small core volume, persistent penumbra volume and favorable outcomes., Competing Interests: The authors have declared that no competing interests exist., (Copyright: © 2024 Tomari et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.)

Background: Incorporating cardiac CT with hyperacute stroke imaging may increase the yield for cardioembolic sources. It is not clarified whether stroke severity influences on rates of intracardiac thrombus. We aimed to investigate a National Institutes of Health Stroke Scale (NIHSS) threshold below which acute cardiac CT was unnecessary., Methods: Consecutive patients with suspected stroke who underwent multimodal brain imaging and concurrent non-gated cardiac CT with delayed timing were prospectively recruited from 1 st December 2020 to 30 th November 2021. We performed receiver operating characteristics analysis of the NIHSS and intracardiac thrombus on hyperacute cardiac CT., Results: A total of 314 patients were assessed (median age 69 years, 61% male). Final diagnoses were ischemic stroke (n=205; 132 etiology-confirmed stroke, independent of cardiac CT and 73 cryptogenic), transient ischemic attack (TIA) (n=21) and stroke-mimic syndromes (n=88). The total yield of cardiac CT was 8 intracardiac thrombus and 1 dissection. Cardiac CT identified an intracardiac thrombus in 6 (4.5%) with etiology-confirmed stroke, 2 (2.7%) with cryptogenic stroke, and none in patients with TIA or stroke-mimic. All of those with intracardiac thrombus had NIHSS ≥4 and this was the threshold below which hyperacute cardiac CT was not justified (sensitivity 100%, specificity 38%, positive predictive value 4.0%, negative predictive value 100%)., Conclusions: A cutoff NIHSS ≥4 may be useful to stratify patients for cardiac CT in the hyperacute stroke setting to optimize its diagnostic yield and reduce additional radiation exposure., Competing Interests: Declaration of Competing Interest The authors have no competing interests to declare that are relevant to the content of this article., (Copyright © 2023. Published by Elsevier Inc.)

Introduction: Evidence-based blood pressure (BP) targets in acute ischaemic stroke are lacking. Previous observational studies have focused on single baseline BP and clinical outcomes, without consideration for dynamic changes. We aim to determine the association between BP parameters including variability, peak, nadir, median and mean during stroke and infarct growth (primary outcome), risk of haemorrhagic transformation, and functional outcome (secondary outcomes)., Methods: Suspected stroke patients were prospectively recruited from a single comprehensive stroke centre. Multimodal computed tomography imaging was used to define infarct core. BP was recorded as per national stroke guidelines during the initial 24 h. Infarct growth and evidence of parenchymal haemorrhage were determined by follow-up magnetic resonance imaging at 24 h. Functional outcome at 3 months was assessed using the modified Rankin Scale. Subgroup analysis was performed according to stroke aetiology and treatment for the association between BP, infarct volume growth, and risk of haemorrhagic transformation. The association between BP parameters and outcomes were determined using regression modelling., Results: A total of 229 patients were included in this study. The median age was 67.4, 64.4% were male, and the baseline National Institutes of Health Stroke Scale was 8. BP variability (BPV) was independently associated with increased infarct growth (multivariate coefficient 1.60, 95% CI: 0.27-2.94, p = 0.19) and an increased odds of parenchymal haemorrhage (adjusted OR 1.21, 95% CI: 1.02-1.44, p = 0.028). The odds of a favourable outcome at 90 days were inversely associated with BPV on simple, but not adjusted logistic regression. On subgroup analysis, only in patients with large vessel occlusions, undergoing endovascular clot retrieval, was BPV associated with infarct growth (multivariate-adjusted coefficient 2.62, 95% CI: 0.53-4.70, p = 0.014) and an increased odds of haemorrhagic transformation (adjusted OR 1.26, 95% CI: 1.01-1.57, p = 0.045)., Conclusion: An increase in BPV was associated with infarct expansion, increased risk of haemorrhagic transformation and was negatively associated with favourable functional outcomes at 3 months., (© 2023 S. Karger AG, Basel.)

Neuroscience has long been an essential driver of progress in artificial intelligence (AI). We propose that to accelerate progress in AI, we must invest in fundamental research in NeuroAI. A core component of this is the embodied Turing test, which challenges AI animal models to interact with the sensorimotor world at skill levels akin to their living counterparts. The embodied Turing test shifts the focus from those capabilities like game playing and language that are especially well-developed or uniquely human to those capabilities - inherited from over 500 million years of evolution - that are shared with all animals. Building models that can pass the embodied Turing test will provide a roadmap for the next generation of AI., (© 2023. The Author(s).)

Background: Failure to reperfuse a cerebral occlusion resulting in a persistent penumbral pattern has not been fully described., Methods: We retrospectively reviewed patients with anterior large vessel occlusion who did not receive reperfusion, and underwent repeated perfusion imaging, with baseline imaging < 6 h after onset and follow-up scans from 16-168 h. A persistent target mismatch (PTM) was defined as core volume of < 100 mL, mismatch ratio > 1.2, and mismatch volume > 10 mL on follow-up imaging. Patients were divided into PTM or non-PTM groups. Ischemic core and penumbral volumes were compared between baseline and follow-up imaging between the two groups, and collateral flow status assessed using CT perfusion collateral index., Results: A total of 25 patients (14 PTM and 11 non-PTM) were enrolled in the study. Median core volumes increased slightly in the PTM group, from 22 to 36 ml. There was a much greater increase in the non-PTM group, from 57 to 190 ml. Penumbral volumes were stable in the PTM group from a median of 79 ml at baseline to 88 ml at follow-up, whereas penumbra was reduced in the non-PTM group, from 120 to 0 ml. Collateral flow status was also better in the PTM group and the median collateral index was 33% compared with 44% in the non-PTM group (p = 0.043)., Conclusion: Multiple patients were identified with limited core growth and large penumbra (persistent target mismatch) > 16 h after stroke onset, likely due to more favorable collateral flow., (© 2022. The Author(s).)

Objective: The Hunter-8 prehospital stroke scale predicts large vessel occlusion in hyperacute ischemic stroke patients (LVO) at hospital admission. We wished to test its performance in the hands of paramedics as part of a prehospital triage algorithm. We aimed to determine (a) the proportion of patients identified by the Hunter-8 algorithm, receiving reperfusion therapies, (b) whether a call to stroke team improved this, and (c) performance for LVO detection using an expanded LVO definition., Methods: A prehospital workflow combining pre-morbid functional status, time from symptom onset, and the Hunter-8 scale was implemented from July 2019. A telephone call to the stroke team was prompted for potential treatment candidates. Classic LVO was defined as a proximal middle cerebral artery (MCA-M1), terminal internal carotid artery, or tandem occlusion. Extended LVO added proximal MCA-M2 and basilar occlusions., Results: From July 2019 to April 2021, there were 363 Hunter-8 activations, 320 analyzed: 181 (56.6%) had confirmed ischemic strokes, 13 (4.1%) transient ischemic attack, 91 (28.5%) stroke mimics, and 35 (10.9%) intracranial hemorrhage. Fifty-two patients (16.3%) received reperfusion therapies, 35 with Hunter-8 ≥ 8. The stroke doctor changed the final destination for 76 patients (23.7%), and five received reperfusion therapies. The AUCs for classic and extended LVO were 0.73 (95% CI 0.66-0.79) and 0.72 (95% CI 0.65-0.77), respectively., Conclusion: The Hunter-8 workflow resulted in 28.7% of confirmed ischemic stroke patients receiving reperfusion therapies, with no secondary transfers to the comprehensive stroke center. The role of communication with stroke team needs to be further explored.

Introduction: Continuous surveillance of stroke admissions has been conducted in the Hunter region, Australia, over the past two decades. We aimed to describe the trends in incidence rates of hospitalised stroke and case-fatality rates in this region, 2001-2019., Methods: From a hospital-based stroke registry, data for admitted adult stroke patients residing in the Hunter region were collected using ICD-10 codes for ischemic and haemorrhagic stroke. Negative binomial regression and logistic regression analysis were used to analyse trends for age-standardised and age-specific incidence rates of hospitalised stroke and 28-day case-fatality rates., Results: A total of 14,662 hospitalisations for stroke in 13,242 individuals were registered. The age-standardised incidence rate declined from 123 per 100,000 population in the 2001-2005 epoch to 96 in the 2016-2019 epoch (mean annual change -2.0%, incidence rate ratio (IRR) = 0.980 [95%CI: 0.976-0.984]). Age-specific analyses identified significant reduction in the group aged 75-84 (1039 per 100,000 population in 2001-2005 to 633 in 2016-2019, annual change -3.5%, IRR= 0.965 [95%CI: 0.960-0.970]). The 28-day case-fatality rates fluctuated over time (18.5% in 2001-2005, 20.8% in 2010-2015, and 17.8% in 2016-2019).  Projected population aging suggests annual volume of patients with new stroke will increase by 77% by 2041 if incidence rates remain unchanged at the 2016-2019 level., Conclusion: Although age-standardised hospitalised stroke incidence rates have declined in the Hunter region, the health system will face an increase in stroke hospitalisations related to the aging population., Competing Interests: Conflict of Interest The author(s) declared there is no potential conflicts of interest with publication of this article., (Copyright © 2022 The Authors. Published by Elsevier Inc. All rights reserved.)

Introduction: A telestroke network in Northern New South Wales, Australia has been developed since 2017. We theorized that the telestroke network development would drive a progressive improvement in stroke care metrics over time. Aim: This study aimed to describe changes in acute stroke workflow metrics over time to determine whether they improved with network experience. Methods: We prospectively collected data of patients assessed by telestroke who received multimodal computed tomography (mCT) and were diagnosed with ischemic stroke or transient ischemic attack from January 2017 to July 2019. The period was divided into two phases (phase 1: January 2017 - October 2018 and phase 2: November 2018 - July 2019). We compared median door-to-call, door-to-image, and door-to-decision time between the two phases. Results: We included 433 patients (243 in phase 1 and 190 in phase 2). Each spoke site treated 1.5-5.2 patients per month. There were Door-to-call time (median 39 in phase 1, 35 min in phase 2, p = 0.18), and door-to-decision time (median 81.5 vs. 83 min, p = 0.31) were not improved significantly. Similarly, in the reperfusion therapy subgroup, door-to-call time (median 29 vs. 24.5 min, p = 0.12) and door-to-decision time (median 70.5 vs. 67.5 min, p = 0.75) remained substantially unchanged. Regression analysis showed no association between time in the network and door-to-decision time (coefficient 1.5, p = 0.32). Conclusion: In our telestroke network, acute stroke timing metrics did not improve over time. There is the need for targeted education and training focusing on both stroke reperfusion competencies and the technical aspects of telestroke in areas with limited workforce and high turnover., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2021 Kashida, Garcia-Esperon, Lillicrap, Miteff, Garcia-Bermejo, Gangadharan, Chew, O'Brien, Evans, Alanati, Bivard, Parsons, Majersik, Spratt, Levi and the members of Northern NSW Telestroke investigators for this project.)

Background and Purpose: Lacunar syndromes correlate with a lacunar stroke on imaging in 50% to 60% of cases. Computed tomography perfusion (CTP) is becoming the preferred imaging modality for acute stroke triage. We aimed to estimate the sensitivity, specificity, and predictive values for noncontrast computed tomography and CTP in lacunar syndromes, and for cortical, subcortical, and posterior fossa regions., Methods: A retrospective analysis of confirmed ischemic stroke patients who underwent acute CTP and follow-up magnetic resonance imaging between 2010 and 2018 was performed. Brain noncontrast computed tomography and CTP were assessed independently by 2 stroke neurologists. Receiver operating characteristic curve analysis was performed to estimate sensitivity, specificity, and area under the curve (AUC) for the detection of strokes in patients with lacunar syndromes using different CTP maps., Results: We found 106 clinical lacunar syndromes, but on diffusion-weighted imaging, these consisted of 59 lacunar, 33 cortical, and 14 posterior fossa strokes. The discrimination of ischemia identification was very poor using noncontrast computed tomography in all 3 regions, but good for cortical (AUC, 0.82) and poor for subcortical and posterior regions (AUCs, 0.55 and 0.66) using automated core-penumbra maps. The addition of delay time and mean transient time maps substantially increased subcortical (AUC, 0.80) and slightly posterior stroke detection (AUC, 0.69)., Conclusions: Analysis of mean transient time and delay time maps in combination with core-penumbra maps improves detection of subcortical and posterior strokes.

Constructing agents with planning capabilities has long been one of the main challenges in the pursuit of artificial intelligence. Tree-based planning methods have enjoyed huge success in challenging domains, such as chess 1 and Go 2 , where a perfect simulator is available. However, in real-world problems, the dynamics governing the environment are often complex and unknown. Here we present the MuZero algorithm, which, by combining a tree-based search with a learned model, achieves superhuman performance in a range of challenging and visually complex domains, without any knowledge of their underlying dynamics. The MuZero algorithm learns an iterable model that produces predictions relevant to planning: the action-selection policy, the value function and the reward. When evaluated on 57 different Atari games 3 -the canonical video game environment for testing artificial intelligence techniques, in which model-based planning approaches have historically struggled 4 -the MuZero algorithm achieved state-of-the-art performance. When evaluated on Go, chess and shogi-canonical environments for high-performance planning-the MuZero algorithm matched, without any knowledge of the game dynamics, the superhuman performance of the AlphaZero algorithm 5 that was supplied with the rules of the game.

Aims: Multimodal computed tomography (mCT) (non-contrast CT, CT angiography, and CT perfusion) is not routinely used to assess posterior fossa strokes. We described the area under the curve (AUC) of brain NCCT, WB-CTP automated core-penumbra maps and comprehensive CTP analysis (automated core-penumbra maps and all perfusion maps) for posterior fossa strokes. Methods: We included consecutive patients with signs and symptoms of posterior fossa stroke who underwent acute mCT and follow up magnetic resonance diffusion weighted imaging (DWI). Multimodal CT images were reviewed blindly and independently by two stroke neurologists and area under the receiver operating characteristic curve (AUC) was used to compare imaging modalities. Results: From January 2014 to December 2019, 83 patients presented with symptoms suggestive of posterior fossa strokes and had complete imaging suitable for inclusion (49 posterior fossa strokes and 34 DWI negative patients). For posterior fossa strokes, comprehensive CTP analysis had an AUC of 0.68 vs. 0.62 for automated core-penumbra maps and 0.55 for NCCT. For cerebellar lesions >5 mL, the AUC was 0.87, 0.81, and 0.66, respectively. Conclusion: Comprehensive CTP analysis increases the detection of posterior fossa lesions compared to NCCT and should be implemented as part of the routine imaging assessment in posterior fossa strokes., (Copyright © 2020 Ostman, Garcia-Esperon, Lillicrap, Tomari, Holliday, Levi, Bivard, Parsons and Spratt.)

Rationale: More than half of patients who receive thrombolysis for acute ischaemic stroke fail to recanalize. Elucidating biological factors which predict recanalization could identify therapeutic targets for increasing thrombolysis success. Hypothesis: We hypothesize that individual patient plasmin potential, as measured by in vitro response to recombinant tissue-type plasminogen activator (rt-PA), is a biomarker of rt-PA response, and that patients with greater plasmin response are more likely to recanalize early. Methods: This study will use historical samples from the Barcelona Stroke Thrombolysis Biobank, comprised of 350 pre-thrombolysis plasma samples from ischaemic stroke patients who received serial transcranial-Doppler (TCD) measurements before and after thrombolysis. The plasmin potential of each patient will be measured using the level of plasmin-antiplasmin complex (PAP) generated after in-vitro addition of rt-PA. Levels of antiplasmin, plasminogen, t-PA activity, and PAI-1 activity will also be determined. Association between plasmin potential variables and time to recanalization [assessed on serial TCD using the thrombolysis in brain ischemia (TIBI) score] will be assessed using Cox proportional hazards models, adjusted for potential confounders. Outcomes: The primary outcome will be time to recanalization detected by TCD (defined as TIBI ≥4). Secondary outcomes will be recanalization within 6-h and recanalization and/or haemorrhagic transformation at 24-h. This analysis will utilize an expanded cohort including ~120 patients from the Targeting Optimal Thrombolysis Outcomes (TOTO) study. Discussion: If association between proteolytic response to rt-PA and recanalization is confirmed, future clinical treatment may customize thrombolytic therapy to maximize outcomes and minimize adverse effects for individual patients., (Copyright © 2020 Lillicrap, Keragala, Draxler, Chan, Ho, Harman, Niego, Holliday, Levi, Garcia-Esperon, Spratt, Gyawali, Bivard, Parsons, Montaner, Bustamante, Cadenas, Cloud, Maguire, Lincz, Kleinig, Attia, Koblar, Hamilton-Bruce, Choi, Worrall and Medcalf.)

Background: In stroke survivors, post-stroke fatigue predicts dependency in daily living and failure to return to work. Modafinil shows promise as a pharmacotherapy to reduce post-stroke fatigue and related sequelae, e.g., poorer functional and clinical outcomes., Aims: This study explored the cost-effectiveness of modafinil in treating post-stroke fatigue in the Australian context, by determining its incremental cost-effectiveness ratio (ICER) and by simulating the potential cost-savings on a national scale, through a re-analysis of MIDAS trial data., Methods: A post hoc cost-effectiveness analysis was undertaken. Part A: patient-level cost and health effect data (Multidimensional Fatigue Inventory (MFI) scores) were derived from the MIDAS trial and analysis undertaken from a health-system perspective. Part B: a secondary analysis simulated the societal impact of modafinil therapy in terms of national productivity costs., Results: Part A: Mean cost of modafinil treatment was AUD$3.60/day/patient for a minimally clinically important change (10 points) in total MFI fatigue score, i.e., AUD$0.36/day/unit change in fatigue score per patient. For the base case scenario, the ICER of using modafinil (versus placebo) was AUD$131.73 ($90.17 - 248.15, for minimum and maximum costs, respectively). Part B: The potential productivity cost-savings to society were calculated as nearly AUD$467 million over 1 year, and up to $383,471,991,248 over 10 years, from the widespread use of modafinil treatment in the Australian population of working-age stroke-survivors, representing a significant societal benefit., Conclusions: Modafinil is a highly cost-effective treatment for post-stroke fatigue, offering significant productivity gains and potential cost-savings to society from the widespread use of modafinil treatment in the Australian population of working-age stroke-survivors., Competing Interests: Declaration of Competing Interest The authors have no conflicts of interest to declare in relation to this study., (Copyright © 2020 Elsevier Inc. All rights reserved.)

Background and Purpose: Telestroke aims to increase access to endovascular clot retrieval (ECR) for rural areas. There is limited information on transfer workflow for ECR in rural settings. We sought to describe the transfer metrics for ECR in a rural telestroke network with respect to decision making. Methods: A retrospective cohort study was employed on consecutive patients transferred to the comprehensive stroke center (CSC) for ECR in a rural hub-and-spoke telestroke network between April 2013 and October 2019, by road or air. Key time-based metrics were analyzed. Results: Sixty-two patients were included. Mean age was 66 years [standard deviation (SD), 14] and median National Institutes of Health Stroke Scale 13 [interquartile range (IQR), 8-18]. Median rural-hospital-door-to-CSC-door (D2D) was 308 min (IQR, 254-351), of which 68% was spent at rural hospitals [door-in-door-out (DIDO); 214 min; IQR, 171-247]. DIDO was longer for air transfers than road ( P = 0.004), primarily because of a median 87 min greater decision-to-departure time (Decision-DO, P < 0.001). In multiple linear regression analysis, intubation but not thrombolysis was associated with significantly longer DIDO. The distance at which the extra speed of an aircraft made up for the delays involved in booking an aircraft was 299 km from the CSC. Conclusions: DIDO is longer for air retrievals compared with road. Decision-DO represents the most important component of DIDO, being longer for air transfers. Systems for rapid transportation of rural ECR candidates need optimization for best patient outcomes, with decision support seen as a potential tool to achieve this., (Copyright © 2020 Gangadharan, Lillicrap, Miteff, Garcia-Bermejo, Wellings, O'Brien, Evans, Alanati, Levi, Parsons, Bivard, Garcia-Esperon and Spratt.)

Introduction: Intravenous thrombolysis (IVT) with recombinant tissue plasminogen activator (rt-PA) is the only approved pharmacological reperfusion therapy for acute ischaemic stroke. Despite population benefit, IVT is not equally effective in all patients, nor is it without significant risk. Uncertain treatment outcome prediction complicates patient treatment selection. This study will develop and validate predictive algorithms for IVT response, using clinical, radiological and blood-based biomarker measures. A secondary objective is to develop predictive algorithms for endovascular thrombectomy (EVT), which has been proven as an effective reperfusion therapy since study inception., Methods and Analysis: The Targeting Optimal Thrombolysis Outcomes Study is a multicenter prospective cohort study of ischaemic stroke patients treated at participating Australian Stroke Centres with IVT and/or EVT. Patients undergo neuroimaging using multimodal CT or MRI at baseline with repeat neuroimaging 24 hours post-treatment. Baseline and follow-up blood samples are provided for research use. The primary outcome is good functional outcome at 90 days poststroke, defined as a modified Rankin Scale (mRS) Score of 0-2. Secondary outcomes are reperfusion, recanalisation, infarct core growth, change in stroke severity, poor functional outcome, excellent functional outcome and ordinal mRS at 90 days. Primary predictive models will be developed and validated in patients treated only with rt-PA. Models will be built using regression methods and include clinical variables, radiological measures from multimodal neuroimaging and blood-based biomarkers measured by mass spectrometry. Predictive accuracy will be quantified using c-statistics and R 2 . In secondary analyses, models will be developed in patients treated using EVT, with or without prior IVT, reflecting practice changes since original study design., Ethics and Dissemination: Patients, or relatives when patients could not consent, provide written informed consent to participate. This study received approval from the Hunter New England Local Health District Human Research Ethics Committee (reference 14/10/15/4.02). Findings will be disseminated via peer-reviewed publications and conference presentations., Competing Interests: Competing interests: None declared., (© Author(s) (or their employer(s)) 2020. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.)

Aims: Penumbral selection is best-evidence practice for thrombectomy in the 6-24 hour window. Moreover, it helps to identify the best responders to thrombolysis. Multimodal computed tomography (mCT) at the primary centre-including noncontrast CT, CT perfusion, and CT angiography-may enhance reperfusion therapy decision-making. We developed a network with five spoke primary stroke sites and assessed safety, feasibility, and influence of mCT in rural hospitals on decision-making for thrombolysis., Methods: Consecutive patients assessed via telemedicine from April 2013 to June 2018. Clinical outcomes were measured, and decision-making compared using theoretical models for reperfusion therapy applied without mCT guidance. Symptomatic intracranial hemorrhage (sICH) was assessed according to Safe Implementation of Treatments in Stroke Thrombolysis Registry criteria., Results: A total of 334 patients were assessed, 240 received mCT, 58 were thrombolysed (24.2%). The mean age of thrombolysed patients was 70 years, median baseline National Institutes of Health Stroke Scale was 10 (IQR 7-18) and 23 (39.7%) had a large vessel occlusion. 1.7% had sICH and 3.5% parenchymal hematoma. Three months poststroke, 55% were independent, compared with 70% in the non-thrombolysed group., Conclusion: Implementation of CTP in rural centers was feasible and led to high thrombolysis rates with low rates of sICH., (© 2019 The Authors. CNS Neuroscience & Therapeutics Published by John Wiley & Sons Ltd.)

Background: Admission outside normal business hours has been associated with prolonged door-to-treatment times and poorer patient outcomes, the so called "weekend effect. " This is the first examination of the weekend effect in a telestroke service that uses multi-modal computed tomography. Aims: To examine differences in workflow and triage between in-hours and out-of-hours calls to a telestroke service. Methods: All patients assessed using the Northern New South Wales (N-NSW) telestroke service from April 2013 to January 2019 were eligible for inclusion (674 in total; 539 with complete data). The primary outcomes measured were differences between in-hours and out-of-hours in door-to-call-to-decision-to-needle times, differences in the proportion of patients confirmed to have strokes or of patients selected for reperfusion therapies or patients with a modified Rankin Score (mRS ≤ 2) at 90 days. Results: There were no significant differences between in-hours and out-of-hours in any of the measured times, nor in the proportions of patients confirmed to have strokes (67.6 and 69.6%, respectively, p = 0.93); selected for reperfusion therapies (22.7 and 22.6%, respectively, p = 0.56); or independent at 3 months (34.8 and 33.6%, respectively, p = 0.770). There were significant differences in times between individual hospitals, and patient presentation more than 4.5 h after symptom onset was associated with slower times (21 minute delay in door-to-call, p = 0.002 and 22 min delay in door-to-image, p = 0.001). Conclusions: The weekend effect is not evident in the Northern NSW telestroke network experience, though this study did identify some opportunities for improvement in the delivery of acute stroke therapies., (Copyright © 2020 Lillicrap, Pinheiro, Miteff, Garcia-Bermejo, Gangadharan, Wellings, O'Brien, Evans, Alanati, Bivard, Parsons, Levi, Garcia-Esperon and Spratt.)

Background and Purpose- Low left ventricular ejection fraction (LVEF) leads to worse outcomes after stroke. We hypothesized that the arterial input function (AIF) variability on perfusion computed tomography, especially the time between scan onset and end of AIF (SO-EndAIF), would reflect reduction of cardiac output. Methods- Retrospective analysis of consecutive stroke patients, who underwent computed tomography between January 2013 and September 2018, was performed in 2 parts. (1) To determine the correlation between SO-EndAIF and LVEF, all patients with a transthoracic echocardiogram performed ±6 months from the time of stroke were included. LVEF was dichotomized as either normal (≥50%) or decreased (<50%). (2) AIF was compared with hypoperfusion volume, defined as delay time >3 seconds and with clinical outcome measured using 3-month modified Rankin Scale. Results- A total of 732 ischemic stroke patients underwent computed tomography, 231 with transthoracic echocardiogram were included in part (1), 393 with outcome data were included in part (2). In part (1), 193/231 (83.5%) had normal LVEF (median 61%) and 38/231 (16.5%) decreased LVEF (median 39%). The low-LVEF group had significantly prolonged SO-EndAIF compared with normal-LVEF group (mean of 39.7 versus 26 second; P <0.001), and larger hypoperfusion lesions (94.9 versus 37.6 mL; P <0.001). SO-EndAIF time was strongly associated with EF, with an area under the curve of 0.86. Twenty nine seconds was the best threshold to distinguish between normal and impaired EF (area under the curve, 0.77). In part (2), the SO-EndAIF ≥29 second group had larger hypoperfusion volumes (21.8 versus 89.7 mL; P <0.001) and infarct core (12.2 versus 2.3 mL; P <0.0001) and patients with SO-EndAIF ≥29 seconds had fewer excellent or good clinical outcomes (modified Rankin Scale score 0-1; 40% versus 22%; OR, 2.79; P <0.001, modified Rankin Scale score 0-2; 65% versus 35%; OR, 1.41; P =0.033). Conclusions- AIF width correlates with ejection fraction in acute ischemic stroke. A 29-second threshold from scan onset to end of AIF accurately predicts reduced LVEF and identifies patients more likely to have worse outcomes after stroke.

Brette contends that the neural coding metaphor is an invalid basis for theories of what the brain does. Here, we argue that it is an insufficient guide for building an artificial intelligence that learns to accomplish short- and long-term goals in a complex, changing environment.

Humans prolifically engage in mental time travel. We dwell on past actions and experience satisfaction or regret. More than storytelling, these recollections change how we act in the future and endow us with a computationally important ability to link actions and consequences across spans of time, which helps address the problem of long-term credit assignment: the question of how to evaluate the utility of actions within a long-duration behavioral sequence. Existing approaches to credit assignment in AI cannot solve tasks with long delays between actions and consequences. Here, we introduce a paradigm where agents use recall of specific memories to credit past actions, allowing them to solve problems that are intractable for existing algorithms. This paradigm broadens the scope of problems that can be investigated in AI and offers a mechanistic account of behaviors that may inspire models in neuroscience, psychology, and behavioral economics.

Many real-world applications require artificial agents to compete and coordinate with other agents in complex environments. As a stepping stone to this goal, the domain of StarCraft has emerged as an important challenge for artificial intelligence research, owing to its iconic and enduring status among the most difficult professional esports and its relevance to the real world in terms of its raw complexity and multi-agent challenges. Over the course of a decade and numerous competitions 1-3 , the strongest agents have simplified important aspects of the game, utilized superhuman capabilities, or employed hand-crafted sub-systems 4 . Despite these advantages, no previous agent has come close to matching the overall skill of top StarCraft players. We chose to address the challenge of StarCraft using general-purpose learning methods that are in principle applicable to other complex domains: a multi-agent reinforcement learning algorithm that uses data from both human and agent games within a diverse league of continually adapting strategies and counter-strategies, each represented by deep neural networks 5,6 . We evaluated our agent, AlphaStar, in the full game of StarCraft II, through a series of online games against human players. AlphaStar was rated at Grandmaster level for all three StarCraft races and above 99.8% of officially ranked human players.

Post-stroke fatigue has a significant impact on stroke survivors' mental and physical well-being. Our recent clinical trial showed significant reduction of post-stroke fatigue with modafinil treatment, however functional connectivity changes in response to modafinil have not yet been explored in stroke survivors with post-stroke fatigue. Twenty-eight participants (multidimensional fatigue inventory-20 ≥ 60) had MRI scans at baseline, and during modafinil and placebo treatment. Resting-state functional MRI data were obtained, and independent component analysis was used to extract functional networks. Resting-state functional connectivity (rsFC) was examined between baseline, modafinil and placebo treatment using permutation testing with threshold-free cluster enhancement. Overall twenty-eight participants (mean age: 62 ± 14.3, mean baseline MFI-20: 72.3 ± 9.24) were included. During modafinil treatment, increased rsFC was observed in the right hippocampus (p = 0.004, 11 voxels) compared to placebo. This coincided with lower rsFC in the left frontoparietal (inferior parietal lobule, p = 0.023, 13 voxels), somatosensory (primary somatosensory cortex; p = 0.009, 32 voxels) and mesolimbic network (temporal pole, p = 0.016, 35 voxels). In conclusion, modafinil treatment induces significant changes in rsFC in post-stroke fatigue. This modulation of rsFC may relate to a reduction of post-stroke fatigue; however, the relationship between sensory processing, neurotransmitter expression and fatigue requires further exploration.

The game of chess is the longest-studied domain in the history of artificial intelligence. The strongest programs are based on a combination of sophisticated search techniques, domain-specific adaptations, and handcrafted evaluation functions that have been refined by human experts over several decades. By contrast, the AlphaGo Zero program recently achieved superhuman performance in the game of Go by reinforcement learning from self-play. In this paper, we generalize this approach into a single AlphaZero algorithm that can achieve superhuman performance in many challenging games. Starting from random play and given no domain knowledge except the game rules, AlphaZero convincingly defeated a world champion program in the games of chess and shogi (Japanese chess), as well as Go., (Copyright © 2018 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S. Government Works.)

Introduction: The incidence of pituitary dysfunction after severe ischemic stroke is unknown, however given the increasing attention to pituitary dysfunction after neurological injuries such as traumatic brain injury, this may represent a novel area of research in stroke. Methods: We perform an arginine and human growth hormone releasing hormone challenge on ischemic stroke patients within a week of symptom onset. Results: Over the study period, 13 patients were successfully tested within a week of stroke (baseline NIHSS 10, range 7-16). Overall, 9(69%) patients had a poor response, with 7(54%) of these patients meeting the criteria for had human growth hormone deficiency. Other measures of pituitary function were within normal ranges. Conclusion: After major ischemic stroke, low GH levels are common and may play a role in stroke recovery.

Deep neural networks have achieved impressive successes in fields ranging from object recognition to complex games such as Go 1,2 . Navigation, however, remains a substantial challenge for artificial agents, with deep neural networks trained by reinforcement learning 3-5 failing to rival the proficiency of mammalian spatial behaviour, which is underpinned by grid cells in the entorhinal cortex 6 . Grid cells are thought to provide a multi-scale periodic representation that functions as a metric for coding space 7,8 and is critical for integrating self-motion (path integration) 6,7,9 and planning direct trajectories to goals (vector-based navigation) 7,10,11 . Here we set out to leverage the computational functions of grid cells to develop a deep reinforcement learning agent with mammal-like navigational abilities. We first trained a recurrent network to perform path integration, leading to the emergence of representations resembling grid cells, as well as other entorhinal cell types 12 . We then showed that this representation provided an effective basis for an agent to locate goals in challenging, unfamiliar, and changeable environments-optimizing the primary objective of navigation through deep reinforcement learning. The performance of agents endowed with grid-like representations surpassed that of an expert human and comparison agents, with the metric quantities necessary for vector-based navigation derived from grid-like units within the network. Furthermore, grid-like representations enabled agents to conduct shortcut behaviours reminiscent of those performed by mammals. Our findings show that emergent grid-like representations furnish agents with a Euclidean spatial metric and associated vector operations, providing a foundation for proficient navigation. As such, our results support neuroscientific theories that see grid cells as critical for vector-based navigation 7,10,11 , demonstrating that the latter can be combined with path-based strategies to support navigation in challenging environments.

Background and Purpose: Transient ischemic attack (TIA) initiates an ischemic cascade without resulting in frank infarction and, as such, represents a novel model to study the effects of this ischemic cascade and secondary neurodegeneration in humans., Methods: Patients with suspected TIA underwent acute brain perfusion imaging, and those with acute ischemia were enrolled into a prospective observational study. We collected baseline and 90-day magnetic resonance imaging, including MP-RAGE (high-resolution T1 sequence) and cognitive assessment with the Montreal Cognitive Assessment. Brain morphometry and within patient statistical analysis were performed to identify changes between baseline and 90-day imaging and clinical assessments., Results: Fifty patients with TIA with acute perfusion lesions were studied. All patients experienced a decrease in global cortical gray matter ( P =0.005). Patients with anterior circulation TIA (n=31) also had a significant reduction in the volume of the pons ( P <0.001), ipsilesional parietal lobe ( P <0.001), occipital lobe ( P =0.002), frontal lobe ( P <0.001), temporal lobe ( P =0.003), and thalamus ( P =0.016). Patients with an anterior perfusion lesion on acute imaging also had a significant decrease in Montreal Cognitive Assessment between baseline and day 90 ( P =0.027), which may be related to the volume of thalamic atrophy ( R 2 =0.28; P =0.009)., Conclusions: In a prospective observational study, patients with TIA confirmed by acute perfusion imaging experienced a significant reduction in global gray matter and focal structural atrophy related to the area of acute ischemia. The atrophy also resulted in a proportional decreased cognitive performance on the Montreal Cognitive Assessment. Further studies are required to identify the mechanisms of this atrophy., (© 2018 American Heart Association, Inc.)

Ischaemic stroke is a major public health issue in both developed and developing nations. Hypothermia is believed to be neuroprotective in cerebral ischaemia. Conversely, elevated brain temperature is associated with poor outcome after ischaemic stroke. Mechanisms of heat exchange in normally-perfused brain are relatively well understood, but these mechanisms have not been studied as extensively during focal cerebral ischaemia. A finite element model (FEM) of heat exchange during focal ischaemia in the human brain was developed, based on the Pennes bioheat equation. This model incorporated healthy (normally-perfused) brain tissue, tissue that was mildly hypoperfused but not at risk of cell death (referred to as oligaemia), tissue that was hypoperfused and at risk of death but not dead (referred to as penumbra) and tissue that had died as a result of ischaemia (referred to as infarct core). The results of simulations using this model were found to match previous in-vivo temperature data for normally-perfused brain. However, the results did not match what limited data are available for hypoperfused brain tissue, in particular the penumbra, which is the focus of acute neuroprotective treatments such as hypothermia. These results suggest that the assumptions of the Pennes bioheat equation, while valid in the brain under normal circumstances, are not valid during focal ischaemia. Further investigation into the heat exchange profiles that do occur during focal ischaemia may yield results for clinical trials of therapeutic hypothermia.

A long-standing goal of artificial intelligence is an algorithm that learns, tabula rasa, superhuman proficiency in challenging domains. Recently, AlphaGo became the first program to defeat a world champion in the game of Go. The tree search in AlphaGo evaluated positions and selected moves using deep neural networks. These neural networks were trained by supervised learning from human expert moves, and by reinforcement learning from self-play. Here we introduce an algorithm based solely on reinforcement learning, without human data, guidance or domain knowledge beyond game rules. AlphaGo becomes its own teacher: a neural network is trained to predict AlphaGo's own move selections and also the winner of AlphaGo's games. This neural network improves the strength of the tree search, resulting in higher quality move selection and stronger self-play in the next iteration. Starting tabula rasa, our new program AlphaGo Zero achieved superhuman performance, winning 100-0 against the previously published, champion-defeating AlphaGo.

Background and Purpose: This study aimed to assess the efficacy of modafinil, a wakefulness-promoting agent in alleviating post-stroke fatigue ≥3 months after stroke. We hypothesized that 200 mg of modafinil daily for 6 weeks would result in reduced symptoms of fatigue compared with placebo., Methods: This single-center phase 2 trial used a randomized, double-blind, placebo-controlled, crossover design. The key inclusion criterion was a multidimensional fatigue inventory score of ≥60. Patients were randomized to either modafinil or placebo for 6 weeks of therapy, then after a 1 week washout period swapped treatment arms for a second 6 weeks of therapy. The primary outcome was the multidimensional fatigue inventory; secondary outcomes included the Montreal cognitive assessment, the Depression, Anxiety, and Stress Scale (DASS), and the Stroke-Specific Quality of Life (SSQoL) scale. The multidimensional fatigue inventory is a self-administered questionnaire with a range of 0 to 100. Treatment efficacy was assessed using linear regression by estimating within-person, baseline-adjusted differences in mean outcomes after therapy. This trial was registered with the Australian New Zealand Clinical Trials Registry (ACTRN12615000350527)., Results: A total of 232 stroke survivors were screened and 36 were randomized. Participants receiving modafinil reported a significant decrease in fatigue (multidimensional fatigue inventory, -7.38; 95% CI, -21.76 to -2.99; P <0.001) and improved quality of life (SSQoL, 11.81; 95% CI, 2.31 to 21.31; P =0.0148) compared with placebo. Montreal cognitive assessment and DASS were not significantly improved with modafinil therapy during the study period ( P >0.05)., Conclusions: Stroke survivors with nonresolving fatigue reported reduced fatigue and improved quality of life after taking 200 mg daily treatment with modafinil., Clinical Trial Registration: URL: https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=368268. Unique identifier: ACTRN12615000350527., (© 2017 The Authors.)

We agree with Lake and colleagues on their list of "key ingredients" for building human-like intelligence, including the idea that model-based reasoning is essential. However, we favor an approach that centers on one additional ingredient: autonomy. In particular, we aim toward agents that can both build and exploit their own internal models, with minimal human hand engineering. We believe an approach centered on autonomous learning has the greatest chance of success as we scale toward real-world complexity, tackling domains for which ready-made formal models are not available. Here, we survey several important examples of the progress that has been made toward building autonomous agents with human-like abilities, and highlight some outstanding challenges.

Background: Fatigue is a common symptom in stroke survivors for which there is currently no proven therapy. Modafinil is a wakefulness-promoting agent with established benefits in other disease models. We aim to test if modafinil will improve patient's self-reported fatigue scores when compared to placebo and if therapy results in increased quality of life., Methods/design: MIDAS is a phase II, single-centre, prospective, double-blinded, randomised, crossover trial of modafinil for the treatment of persistent fatigue in survivors of ischaemic stroke. The inclusion criteria will require an average score of 12 or more across all domains of the Multi-dimensional Fatigue Inventory (MFI-20) and the diagnosis of a stroke more than 6 months prior. Patients will be randomised 1:1 to receive either modafinil 200 mg daily or placebo for a period of 6 weeks, after which a crossover will occur where patients who are on modafinil will begin taking placebo and vice versa. The primary outcome will be improvement in fatigue as measured by the MFI-20. Secondary outcomes will include changes in the Fatigue Severity Scale, improved cognition measured using the Montreal Cognitive Assessment, improvement in mood as determined by the Depression, Anxiety and Stress Scale and improvement in each patient's stroke-specific quality of life score. All participants will also undergo magnetic resonance imaging (MRI) at baseline, crossover and study conclusion to measure cerebral blood flow on arterial spin labelling and brain activity on resting state functional MRI. This study will comply with the CONSORT guidelines. The projected sample size requirement is 36 participants in a crossover trial giving a power of 80 % and a type-1 error rate of 0.05., Discussion: MIDAS seeks to enhance the quality of life in stroke survivors by assisting or resolving stroke-associated fatigue., Trial Registration: ACTRN12615000350527 , registered on the 17 April 2015. Protocol version 3, approved 16 June 2015.

The game of Go has long been viewed as the most challenging of classic games for artificial intelligence owing to its enormous search space and the difficulty of evaluating board positions and moves. Here we introduce a new approach to computer Go that uses 'value networks' to evaluate board positions and 'policy networks' to select moves. These deep neural networks are trained by a novel combination of supervised learning from human expert games, and reinforcement learning from games of self-play. Without any lookahead search, the neural networks play Go at the level of state-of-the-art Monte Carlo tree search programs that simulate thousands of random games of self-play. We also introduce a new search algorithm that combines Monte Carlo simulation with value and policy networks. Using this search algorithm, our program AlphaGo achieved a 99.8% winning rate against other Go programs, and defeated the human European Go champion by 5 games to 0. This is the first time that a computer program has defeated a human professional player in the full-sized game of Go, a feat previously thought to be at least a decade away.

Recent work on the evolution of behaviour is set in a structured population, providing a systematic way to describe gene flow and behavioural interactions. To obtain analytical results one needs a structure with considerable regularity. Our results apply to such "homogeneous" structures (e.g., lattices, cycles, and island models). This regularity has been formally described by a "node-transitivity" condition but in mathematics, such internal symmetry is powerfully described by the theory of mathematical groups. Here, this theory provides elegant direct arguments for a more general version of a number of existing results. Our main result is that in large "group-structured" populations, primary fitness effects on others play no role in the evolution of the behaviour. The competitive effects of such a trait cancel the primary effects, and the inclusive fitness effect is given by the direct effect of the actor on its own fitness. This result is conditional on a number of assumptions such as (1) whether generations overlap, (2) whether offspring dispersal is symmetric, (3) whether the trait affects fecundity or survival, and (4) whether the underlying group is abelian. We formulate a number of results of this type in finite and infinite populations for both Moran and Wright-Fisher demographies., (© 2010 The Author(s). Evolution© 2010 The Society for the Study of Evolution.)

Social learning (learning through observation or interaction with other individuals) is widespread in nature and is central to the remarkable success of humanity, yet it remains unclear why copying is profitable and how to copy most effectively. To address these questions, we organized a computer tournament in which entrants submitted strategies specifying how to use social learning and its asocial alternative (for example, trial-and-error learning) to acquire adaptive behavior in a complex environment. Most current theory predicts the emergence of mixed strategies that rely on some combination of the two types of learning. In the tournament, however, strategies that relied heavily on social learning were found to be remarkably successful, even when asocial information was no more costly than social information. Social learning proved advantageous because individuals frequently demonstrated the highest-payoff behavior in their repertoire, inadvertently filtering information for copiers. The winning strategy (discountmachine) relied nearly exclusively on social learning and weighted information according to the time since acquisition.

To learn effectively, an adaptive controller needs to know its sensitivity derivatives--the variables that quantify how system performance depends on the commands from the controller. In the case of biological sensorimotor control, no one has explained how those derivatives themselves might be learned, and some authors suggest they are not learned at all but are known innately. Here we show that this knowledge cannot be solely innate, given the adaptive flexibility of neural systems. And we show how it could be learned using forms of information transport that are available in the brain. The mechanism, which we call implicit supervision, helps explain the flexibility and speed of sensorimotor learning and our ability to cope with high-dimensional work spaces and tools.

For double-stranded RNA (dsRNA) to signal the presence of foreign (non-self) nucleic acid, self-RNA-self-RNA interactions should be minimized. Indeed, self-RNAs appear to have been fine-tuned over evolutionary time by the introduction of purines in clusters in the loop regions of stem-loop structures. This adaptation should militate against the "kissing" interactions which initiate formation of dsRNA. Our analyses of virus base compositions suggest that, to avoid triggering the host cell's dsRNA surveillance mechanism, most viruses purine-load their RNAs to resemble host RNAs ("stealth" strategy). However, some GC-rich latent viruses (HTLV-1, EBV) pyrimidine-load their RNAs. It is suggested that when virus production begins, these RNAs suddenly increase in concentration and impair host mRNA function by virtue of an excess of complementary "kissing" interactions ("surprise" strategy). Remarkably, the only mRNA expressed in the most fundamental form of EBV latency (the "EBNA-1 program") is purine-loaded. This apparent stealth strategy is reinforced by a simple sequence repeat which prefers purine-rich codons. During latent infection the EBNA-1 protein may evade recognition by cytotoxic T-cells, not by virtue of containing a simple sequence amino acid repeat as has been proposed, but by virtue of the encoding mRNA being purine-loaded to prevent interactions with host RNAs of either genic or non-genic origin., (Copyright 2001 Academic Press.)