Acropolis Educational Resources Repository
Not a member yet
31278 research outputs found
Sort by
A clinical deep learning framework for continually learning from cardiac signals across diseases, time, modalities, and institutions.
No full textDeep learning algorithms trained on instances that violate the assumption of being independent and identically distributed (i.i.d.) are known to experience destructive interference, a phenomenon characterized by a degradation in performance. Such a violation, however, is ubiquitous in clinical settings where data are streamed temporally from different clinical sites and from a multitude of physiological sensors. To mitigate this interference, we propose a continual learning strategy, entitled CLOPS, that employs a replay buffer. To guide the storage of instances into the buffer, we propose end-to-end trainable parameters, termed task-instance parameters, that quantify the difficulty with which data points are classified by a deep-learning system. We validate the interpretation of these parameters via clinical domain knowledge. To replay instances from the buffer, we exploit uncertainty-based acquisition functions. In three of the four continual learning scenarios, reflecting transitions across diseases, time, data modalities, and healthcare institutions, we show that CLOPS outperforms the state-of-the-art methods, GEM1 and MIR2. We also conduct extensive ablation studies to demonstrate the necessity of the various components of our proposed strategy. Our framework has the potential to pave the way for diagnostic systems that remain robust over time
Transfusion management of severe anaemia in African children: a consensus algorithm
No full textThe phase III Transfusion and Treatment of severe anaemia in African Children Trial (TRACT) found that conservative management of uncomplicated severe anaemia [haemoglobin (Hb) 40–60 g/l] was safe, and that transfusion volume (20 vs. 30 ml/kg whole blood equivalent) for children with severe anaemia (Hb 37·5°C). In 2020 a stakeholder meeting of paediatric and blood transfusion groups from Africa reviewed the results and additional analyses. Among all 3196 children receiving an initial transfusion there was no evidence that nutritional status, presence of shock, malaria parasite burden or sickle cell disease status influenced outcomes or modified the interaction with fever status on volume required. Fever status at the time of ordering blood was a reliable determinant of volume required for optimal outcome. Elevated heart and respiratory rates normalised irrespective of transfusion volume and without diuretics. By consensus, a transfusion management algorithm was developed, incorporating three additional measurements of Hb post-admission, alongside clinical monitoring. The proposed algorithm should help clinicians safely implement findings from TRACT. Further research should assess its implementation in routine clinical practice
Prevalence of posttraumatic stress disorder after infectious disease pandemics in the twenty-first century, including COVID-19: a meta-analysis and systematic review
No full textPandemics have become more frequent and more complex during the twenty-first century. Posttraumatic stress disorder (PTSD) following pandemics is a significant public health concern. We sought to provide a reliable estimate of the worldwide prevalence of PTSD after large-scale pandemics as well as associated risk factors, by a systematic review and meta-analysis. We systematically searched the MedLine, Embase, PsycINFO, Web of Science, CNKI, WanFang, medRxiv, and bioRxiv databases to identify studies that were published from the inception up to August 23, 2020, and reported the prevalence of PTSD after pandemics including sudden acute respiratory syndrome (SARS), H1N1, Poliomyelitis, Ebola, Zika, Nipah, Middle Eastern respiratory syndrome coronavirus (MERS-CoV), H5N1, and coronavirus disease 2019 (COVID-19). A total of 88 studies were included in the analysis, with 77 having prevalence information and 70 having risk factors information. The overall pooled prevalence of post-pandemic PTSD across all populations was 22.6% (95% confidence interval (CI): 19.9-25.4%, I2: 99.7%). Healthcare workers had the highest prevalence of PTSD (26.9%; 95% CI: 20.3-33.6%), followed by infected cases (23.8%: 16.6-31.0%), and the general public (19.3%: 15.3-23.2%). However, the heterogeneity of study findings indicates that results should be interpreted cautiously. Risk factors including individual, family, and societal factors, pandemic-related factors, and specific factors in healthcare workers and patients for post-pandemic PTSD were summarized and discussed in this systematic review. Long-term monitoring and early interventions should be implemented to improve post-pandemic mental health and long-term recovery
Learning to control over unknown wireless channels
No full textEmerging control applications in the Internet-of-Things are increasingly relying on communication networks and wireless channels to close the loop. Traditional model-based approaches, i.e., assuming a known wireless channel model, are focused on analyzing stability and designing appropriate controller structures. Such modeling is challenging as wireless channels are typically unknown a priori and only available through data samples. In this work we aim to design data-based controllers using channel samples and provide high confidence guarantees on the performance of these controllers when deployed over the actual unknown channel. To achieve these results we combine statistical learning (concentration inequalities) with structural properties of our problem (monotonicity with respect to the unknown channel parameters), and also provide sample complexity analysis
Spin-current mediated exchange coupling in MgO-based magnetic tunnel junctions
No full textHeterostructures composed of ferromagnetic layers that are mutually interacting through a nonmagnetic spacer are at the core of magnetic sensor and memory devices. In the present study, layer-resolved ferromagnetic resonance was used to investigate the coupling between the magnetic layers of a Co/MgO/Permalloy magnetic tunnel junction. Two magnetic resonance peaks were observed for both magnetic layers, as probed at the Co and Ni L3 x-ray absorption edges, showing a strong interlayer interaction through the insulating MgO barrier. A theoretical model based on the Landau-Lifshitz-Gilbert-Slonczewski equation was developed, including exchange coupling and spin pumping between the magnetic layers. Fits to the experimental data were carried out, both with and without a spin pumping term, and the goodness of the fit was compared using a likelihood ratio test. This rigorous statistical approach provides an unambiguous proof of the existence of interlayer coupling mediated by spin pumping
Artificial intelligence in local government: enabling artificial intelligence for good governance in UK local authorities
No full textIn this, OxCAIGG’s second major research output of the year, author Thomas Vogl explores the use of artificial intelligence by UK local authorities. While there have been a number of successful projects related to back-office automation, predictive analytics for decision support, or the use of chatbots for interactions with resident, Vogl finds that governments face important practical challenges in successfully adopting AI. This briefing note identifies three key challenges and three key enablers that local authorities face when working with AI
Recent advances in understanding and combatting Neisseria gonorrhoeae: a genomic perspective
No full textThe sexually transmitted infection (STI) gonorrhoea remains a major global public health concern. The World Health Organization (WHO) estimates that 87 million new cases in individuals who were 15 to 49 years of age occurred in 2016. The growing number of gonorrhoea cases is concerning given the rise in gonococci developing antimicrobial resistance (AMR). Therefore, a global action plan is needed to facilitate surveillance. Indeed, the WHO has made surveillance leading to the elimination of STIs (including gonorrhoea) a global health priority. The availability of whole genome sequence data offers new opportunities to combat gonorrhoea. This can be through (i) enhanced surveillance of the global prevalence of AMR, (ii) improved understanding of the population biology of the gonococcus, and (iii) opportunities to mine sequence data in the search for vaccine candidates. Here, we review the current status in Neisseria gonorrhoeae genomics. In particular, we explore how genomics continues to advance our understanding of this complex pathogen
Nature-based solutions can help cool the planet — if we act now
No full textAnalysis suggests that to limit global temperature rise, we must slash emissions and invest now to protect, manage and restore ecosystems and land for the future
Hyperpolarized magnetic resonance shows that the antiischemic drug meldonium leads to increased flux through pyruvate dehydrogenase in vivo resulting in improved postischemic function in the diabetic heart
No full textThe diabetic heart has a decreased ability to metabolize glucose. The anti‐ischemic drug meldonium may provide a route to counteract this by reducing l‐carnitine levels, resulting in improved cardiac glucose utilization. Therefore, the aim of this study was to use the novel technique of hyperpolarized magnetic resonance to investigate the in vivo effects of treatment with meldonium on cardiac metabolism and function in control and diabetic rats. Thirty‐six male Wistar rats were injected either with vehicle, or with streptozotocin (55 mg/kg) to induce a model of type 1 diabetes. Daily treatment with either saline or meldonium (100 mg/kg/day) was undertaken for three weeks. in vivo cardiac function and metabolism were assessed with CINE MRI and hyperpolarized magnetic resonance respectively. Isolated perfused hearts were challenged with low‐flow ischemia/reperfusion to assess the impact of meldonium on post‐ischemic recovery. Meldonium had no significant effect on blood glucose concentrations or on baseline cardiac function. However, hyperpolarized magnetic resonance revealed that meldonium treatment elevated pyruvate dehydrogenase flux by 3.1‐fold and 1.2‐fold in diabetic and control animals, respectively, suggesting an increase in cardiac glucose oxidation. Hyperpolarized magnetic resonance further demonstrated that meldonium reduced the normalized acetylcarnitine signal by 2.1‐fold in both diabetic and control animals. The increase in pyruvate dehydrogenase flux in vivo was accompanied by an improvement in post‐ischemic function ex vivo, as meldonium elevated the rate pressure product by 1.3‐fold and 1.5‐fold in the control and diabetic animals, respectively. In conclusion, meldonium improves in vivo pyruvate dehydrogenase flux in the diabetic heart, contributing to improved cardiac recovery after ischemia