Search

Presentations during the panelsession of the OpenAIRE Week,General Assembly. The panel examined the shared building blocks for Open Science, around policy and infrastructure and identifiedkey takeaways.

This work was presented in COAR 2018 Annual Meeting and General Assembly, Hamburg (Germany)

Accurate rainfall forecasting is critical because it has a great impact on people's social and economic activities. Recent trends on various literatures show that Deep Learning (Neural Network) is a promising methodology to tackle many challenging tasks. In this study, we introduce a brand-new data-driven precipitation prediction model called DeepRain. This model predicts the amount of rainfall from weather radar data, which is three-dimensional and four-channel data, using convolutional LSTM (ConvLSTM). ConvLSTM is a variant of LSTM (Long Short-Term Memory) containing a convolution operation inside the LSTM cell. For the experiment, we used radar reflectivity data for a two-year period whose input is in a time series format in units of 6 min divided into 15 records. The output is the predicted rainfall information for the input data. Experimental results show that two-stacked ConvLSTM reduced RMSE by 23.0% compared to linear regression., Climate Informatics Workshop 2017

Summary Background Information on the natural duration of carbapenem-resistant Enterobacteriaceae (CRE) carriage and factors associated with persistence of carriage is limited. Aim To evaluate the clinical variables associated with persistent carriage of Klebsiella pneumoniae carbapenemase (KPC)-producing CRE. Methods Data for patients admitted between June 2015 and December 2016 who were identified as KPC-producing CRE carriers by either rectal swabs or clinical cultures were reviewed retrospectively. Patients with follow-up culture data for three months after initial acquisition were included. Regression models were used to evaluate the clinical variables associated with persistence of carriage. Findings Of the 100 eligible patients, 50 patients (50%) experienced spontaneous decolonization within three months. Among the 50 patients (50%) who remained culture positive after three months, 26 patients carried KPC-producing CRE after six months. Multi-variable analysis revealed that re-admission [adjusted odds ratio (aOR) 9.96; 95% confidence interval (CI) 1.13–87.98; P=0.039], duration of hospitalization (aOR 1.03; 95% CI 1.01–1.05; P=0.003), positive clinical culture (aOR 6.26; 95% CI 1.28–30.54; P=0.023) and carbapenem use (OR 9.15; 95% CI 1.85–45.27; P=0.007) were predictive for persistent carriage after six months. Conclusion The results suggest that patients with KPC-producing CRE in clinical specimens who are using carbapenem, particularly those with multiple and prolonged hospitalizations, are more likely to remain carriers after six months of initial acquisition. This information is useful for coordinating strategies for pre-emptive isolation by predicting the CRE carriage status appropriately, and ensuring active surveillance through risk factor stratification.

Background The purpose of the present study is to review the international literature, using a systematic review, for studies comparing propofol and dexmedetomidine for drug-induced sleep endoscopy (DISE) or sedation in which there is a description of the effect of the agents on the upper airway and associated variables (e.g., vital signs, sedation scores). Methods This is a systematic review through October 4, 2016. PubMed/MEDLINE and four additional databases were accessed for this study. Results Two hundred twenty studies were screened. 79 were downloaded, and 10 met criteria. The majority of the studies identified dexmedetomidine as the preferred pharmacologic agent for DISE due to an overall safer and more stable profile based upon hemodynamic stability. However, propofol provided greater airway obstruction with oxygen desaturations. With either agent, the degree of obstruction in the upper airway lacks some degree of validity as to whether the obstructions accurately represent natural sleep or are simply a drug induced effect. Conclusion Dexmedetomidine and propofol have their advantages and disadvantages during DISE. Generally, dexmedetomidine was preferred and seemed to provide a more stable profile based upon cardiopulmonary status. However, propofol has a quicker onset, has a shorter half-life, and can demonstrate larger degrees of obstruction, which might more accurately reflect what happens during REM sleep. Additional research is recommended.

Predicting unseen weather phenomena is an important issue for disaster management. In this paper, we suggest a model for a convolutional sequence-to-sequence autoencoder for predicting undiscovered weather situations from previous satellite images. We also propose a symmetric skip connection between encoder and decoder modules to produce more comprehensive image predictions. To examine our model performance, we conducted experiments for each suggested model to predict future satellite images from historical satellite images. A specific combination of skip connection and sequence-to-sequence autoencoder was able to generate closest prediction from the ground truth image., Comment: Workshop on Deep Learning for Physical Sciences (DLPS 2017), NIPS 2017, Long Beach, CA, USA

Advances in remote sensing technologies have made it possible to use high-resolution visual data for weather observation and forecasting tasks. We propose the use of multi-layer neural networks for understanding complex atmospheric dynamics based on multichannel satellite images. The capability of our model was evaluated by using a linear regression task for single typhoon coordinates prediction. A specific combination of models and different activation policies enabled us to obtain an interesting prediction result in the northeastern hemisphere (ENH)., Comment: Under review as a workshop paper at CI 2017

In any search-based digital library (DL) systems dealing with a non-trivial number of documents, users are often required to go through a long list of short document descriptions in order to identify what they are looking for. To tackle the problem, a variety of document organization algorithms and/or visualization techniques have been used to guide users in selecting relevant documents. Since these techniques require heavy computations, however, we developed a presentation server designed to serve as an intermediary between retrieval servers and clients equipped with a visualization interface. In addition, we designed our own visual interface by which users can view a set of documents from different perspectives through layers of document maps. We finally ran experiments to show that the visual interface, in conjunction with the presentation server, indeed helps users in selecting relevant documents from the retrieval results., 13 pages