CORA logo

CORA

Cork Open Research Archive (CORA) is UCC’s Open Access institutional repository which enables UCC researchers to make their research outputs freely available and accessible.

 

UCC Research Communities

Recent Submissions

Item
The Ruthwell Cross and Irish High Crosses: Some points of comparison and contrast
(Royal Irish Academy, 1985-10) Ó Carragáin, Éamonn
Item
Bacteriophage-host interactions as a platform to establish the role of phages in modulating the microbial composition of fermented foods
(OAE Publishing, 2022) White, Kelsey; Yu, Jun-Hyeok; Eraclio, Giovanni; Dal Bello, Fabio; Nauta, Arjen; Mahony, Jennifer; van Sinderen, Douwe; Science Foundation Ireland; Irish Research Council
Food fermentation relies on the activity of robust starter cultures, which are commonly comprised of lactic acid bacteria such as Lactococcus and Streptococcus thermophilus. While bacteriophage infection represents a persistent threat that may cause slowed or failed fermentations, their beneficial role in fermentations is also being appreciated. In order to develop robust starter cultures, it is important to understand how phages interact with and modulate the compositional landscape of these complex microbial communities. Both culture-dependent and -independent methods have been instrumental in defining individual phage-host interactions of many lactic acid bacteria (LAB). This knowledge needs to be integrated and expanded to obtain a full understanding of the overall complexity of such interactions pertinent to fermented foods through a combination of culturomics, metagenomics, and phageomics. With such knowledge, it is believed that factory-specific detection and monitoring systems may be developed to ensure robust and reliable fermentation practices. In this review, we explore/discuss phage-host interactions of LAB, the role of both virulent and temperate phages on the microbial composition, and the current knowledge of phageomes of fermented foods.
Item
Neonatal hypoxic-ischemic encephalopathy grading from multi-channel EEG time-series data using a fully convolutional neural network
(MDPI, 2023-10-25) Yu, Shuwen; Marnane, William P.; Boylan, Geraldine B.; Lightbody, Gordon; Science Foundation Ireland; Wellcome Trust
A deep learning classifier is proposed for grading hypoxic-ischemic encephalopathy (HIE) in neonates. Rather than using handcrafted features, this architecture can be fed with raw EEG. Fully convolutional layers were adopted both in the feature extraction and classification blocks, which makes this architecture simpler, and deeper, but with fewer parameters. Here, two large (335 h and 338 h, respectively) multi-center neonatal continuous EEG datasets were used for training and testing. The model was trained based on weak labels and channel independence. A majority vote method was used for the post-processing of the classifier results (across time and channels) to increase the robustness of the prediction. A dimension reduction tool, UMAP, was used to visualize the model classification effect. The proposed system achieved an accuracy of 86.09% (95% confidence interval: 82.41–89.78%), an MCC of 0.7691, and an AUC of 86.23% on the large unseen test set. Two convolutional neural network architectures which utilized time-frequency distribution features were selected as the baseline as they had been developed or tested on the same datasets. A relative improvement of 23.65% in test accuracy was obtained as compared with the best baseline. In addition, if only one channel was available, the test accuracy was only reduced by 2.63–5.91% compared with making decisions based on the eight channels.
Item
Neonatal hypoxic-ischemic encephalopathy grading from multi-channel EEG time-series data using a fully convolutional neural network
(Institute of Electrical and Electronics Engineers (IEEE), 2023-10-18) Yu, Shuwen; Marnane, William P.; Boylan, Geraldine B.; Lightbody, Gordon; Science Foundation Ireland; Wellcome Trust
A deep learning classifier is proposed for hypoxic-ischemic encephalopathy (HIE) grading in neonates. Rather than using any features, this architecture can be fed with raw EEG. Fully convolutional layers were adopted both in the feature extraction and classification blocks, which makes this architecture simpler, and deeper, but with fewer parameters. Here two large (335h and 338h respectively) multi-center neonatal continuous EEG datasets were used for training and test. The model was trained based on weak labels and channel independence. A majority vote method was used for the post-processing of the classifier results (across time and channels) to increase the robustness of the prediction. The proposed system achieved an accuracy of 86.09% (95% confidence interval: 82.41% ∼89.78%), an MCC of 0.7691, and an AUC of 86.23% on the large unseen test set. Two convolutional neural network architectures which utilized time-frequency distribution features were selected as the baseline as they had been developed or tested on the same datasets. A relative improvement of 23.65% in test accuracy was obtained as compared with the best baseline.