Short-term Memory Representations Research Articles

Improving Healthcare Communication: AI-Driven Emotion Classification in Imbalanced Patient Text Data with Explainable Models

Sentiment analysis is crucial in healthcare to understand patients’ emotions, automatically identifying the feelings of patients suffering from serious illnesses (cancer, AIDS, or Ebola) with an artificial intelligence model that constitutes a major challenge to help health professionals. This study presents a comparative study on different machine learning (logistic regression, naive Bayes, and LightGBM) and deep learning models: long short-term memory (LSTM) and bidirectional encoder representations from transformers (BERT) for classify health feelings thanks to textual data related to patients with serious illnesses. Considering the class imbalance of the dataset, various resampling techniques are investigated. The approach is complemented by an explainable model, LIME, to understand the shortcomings of the classification results. The results highlight the superior performance of the BERT and LSTM models with an F1-score of 89%.

An Environmental Uncertainty Perception Framework for Misinformation Detection and Spread Prediction in the COVID-19 Pandemic: Artificial Intelligence Approach.

Amidst the COVID-19 pandemic, misinformation on social media has posed significant threats to public health. Detecting and predicting the spread of misinformation are crucial for mitigating its adverse effects. However, prevailing frameworks for these tasks have predominantly focused on post-level signals of misinformation, neglecting features of the broader information environment where misinformation originates and proliferates. This study aims to create a novel framework that integrates the uncertainty of the information environment into misinformation features, with the goal of enhancing the model's accuracy in tasks such as misinformation detection and predicting the scale of dissemination. The objective is to provide better support for online governance efforts during health crises. In this study, we embraced uncertainty features within the information environment and introduced a novel Environmental Uncertainty Perception (EUP) framework for the detection of misinformation and the prediction of its spread on social media. The framework encompasses uncertainty at 4 scales of the information environment: physical environment, macro-media environment, micro-communicative environment, and message framing. We assessed the effectiveness of the EUP using real-world COVID-19 misinformation data sets. The experimental results demonstrated that the EUP alone achieved notably good performance, with detection accuracy at 0.753 and prediction accuracy at 0.71. These results were comparable to state-of-the-art baseline models such as bidirectional long short-term memory (BiLSTM; detection accuracy 0.733 and prediction accuracy 0.707) and bidirectional encoder representations from transformers (BERT; detection accuracy 0.755 and prediction accuracy 0.728). Additionally, when the baseline models collaborated with the EUP, they exhibited improved accuracy by an average of 1.98% for the misinformation detection and 2.4% for spread-prediction tasks. On unbalanced data sets, the EUP yielded relative improvements of 21.5% and 5.7% in macro-F1-score and area under the curve, respectively. This study makes a significant contribution to the literature by recognizing uncertainty features within information environments as a crucial factor for improving misinformation detection and spread-prediction algorithms during the pandemic. The research elaborates on the complexities of uncertain information environments for misinformation across 4 distinct scales, including the physical environment, macro-media environment, micro-communicative environment, and message framing. The findings underscore the effectiveness of incorporating uncertainty into misinformation detection and spread prediction, providing an interdisciplinary and easily implementable framework for the field.

What can we learn from a Chinese social media used by glaucoma patients?

PurposeOur study aims to discuss glaucoma patients’ needs and Internet habits using big data analysis and Natural Language Processing (NLP) based on deep learning (DL).MethodsIn this retrospective study, we used web crawler technology to crawl glaucoma-related topic posts from the glaucoma bar of Baidu Tieba, China. According to the contents of topic posts, we classified them into posts with seeking medical advice and without seeking medical advice (social support, expressing emotions, sharing knowledge, and others). Word Cloud and frequency statistics were used to analyze the contents and visualize the keywords of topic posts. Two DL models, Bidirectional Long Short-Term Memory (Bi-LSTM) and Bidirectional Encoder Representations from Transformers (BERT), were trained to identify the posts seeking medical advice. The evaluation matrices included: accuracy, F1 value, and the area under the ROC curve (AUC).ResultsA total of 10,892 topic posts were included, among them, most were seeking medical advice (N = 7071, 64.91%), and seeking advice regarding symptoms or examination (N = 4913, 45.11%) dominated the majority. The following were searching for social support (N = 2362, 21.69%), expressing emotions (N = 497, 4.56%), and sharing knowledge (N = 527, 4.84%) in sequence. The word cloud analysis results showed that ocular pressure, visual field, examination, and operation were the most frequent words. The accuracy, F1 score, and AUC were 0.891, 0.891, and 0.931 for the BERT model, 0.82, 0.821, and 0.890 for the Bi-LSTM model.ConclusionSocial media can help enhance the patient-doctor relationship by providing patients’ concerns and cognition about glaucoma in China. NLP can be a powerful tool to reflect patients’ focus on diseases. DL models performed well in classifying Chinese medical-related texts, which could play an important role in public health monitoring.

Hybrid Feature Extraction for Multi-Label Emotion Classification in English Text Messages

Emotions are vital for identifying an individual’s attitude and mental condition. Detecting and classifying emotions in Natural Language Processing applications can improve Human–Computer Interaction systems, leading to effective decision making in organizations. Several studies on emotion classification have employed word embedding as a feature extraction method, but they do not consider the sentiment polarity of words. Moreover, relying exclusively on deep learning models to extract linguistic features may result in misclassifications due to the small training dataset. In this paper, we present a hybrid feature extraction model using human-engineered features combined with deep learning based features for emotion classification in English text. The proposed model uses data augmentation, captures contextual information, integrates knowledge from lexical resources, and employs deep learning models, including Bidirectional Long Short-Term Memory (Bi-LSTM) and Bidirectional Encoder Representation and Transformer (BERT), to address the issues mentioned above. The proposed model with hybrid features attained the highest Jaccard accuracy on two of the benchmark datasets, with 68.40% on SemEval-2018 and 53.45% on the GoEmotions dataset. The results show the significance of the proposed technique, and we can conclude that the incorporation of the hybrid features improves the performance of the baseline models.

A transformer-based deep learning model for Persian moral sentiment analysis

Moral expressions in online communications can have a serious impact on framing discussions and subsequent online behaviours. Despite research on extracting moral sentiment from English text, other low-resource languages, such as Persian, lack enough resources and research about this important topic. We address this issue using the Moral Foundation theory (MFT) as the theoretical moral psychology paradigm. We developed a Twitter data set of 8000 tweets that are manually annotated for moral foundations and also we established a baseline for computing moral sentiment from Persian text. We evaluate a plethora of state-of-the-art machine learning models, both rule-based and neural, including distributed dictionary representation (DDR), long short-term memory (LSTM) and bidirectional encoder representations from transformer (BERT). Our findings show that among different models, fine-tuning a pre-trained Persian BERT language model with a linear network as the classifier yields the best results. Furthermore, we analysed this model to find out which layer of the model contributes most to this superior accuracy. We also proposed an alternative transformer-based model that yields competitive results to the BERT model despite its lower size and faster inference time. The proposed model can be used as a tool for analysing moral sentiment and framing in Persian texts for downstream social and psychological studies. We also hope our work provides some resources for further enhancing the methods for computing moral sentiment in Persian text.

Similarity-Based Interference in Sentence Comprehension in Aphasia: a Computational Evaluation of Two Models of Cue-Based Retrieval

Sentence comprehension requires the listener to link incoming words with short-term memory representations in order to build linguistic dependencies. The cue-based retrieval theory of sentence processing predicts that the retrieval of these memory representations is affected by similarity-based interference. We present the first large-scale computational evaluation of interference effects in two models of sentence processing — the activation-based model and a modification of the direct-access model — in individuals with aphasia (IWA) and control participants in German. The parameters of the models are linked to prominent theories of processing deficits in aphasia, and the models are tested against two linguistic constructions in German: pronoun resolution and relative clauses. The data come from a visual-world eye-tracking experiment combined with a sentence-picture matching task. The results show that both control participants and IWA are susceptible to retrieval interference, and that a combination of theoretical explanations (intermittent deficiencies, slow syntax, and resource reduction) can explain IWA’s deficits in sentence processing. Model comparisons reveal that both models have a similar predictive performance in pronoun resolution, but the activation-based model outperforms the direct-access model in relative clauses.

Predicting song genre with deep learning

Purpose Genre classification of songs based on lyrics is a challenging task even for humans, however, state-of-the-art natural language processing has recently offered advanced solutions to this task. The purpose of this study is to advance the understanding and application of natural language processing and deep learning in the domain of music genre classification, while also contributing to the broader themes of global knowledge and communication, and sustainable preservation of cultural heritage. Design/methodology/approach The main contribution of this study is the development and evaluation of various machine and deep learning models for song genre classification. Additionally, we investigated the effect of different word embeddings, including Global Vectors for Word Representation (GloVe) and Word2Vec, on the classification performance. The tested models range from benchmarks such as logistic regression, support vector machine and random forest, to more complex neural network architectures and transformer-based models, such as recurrent neural network, long short-term memory, bidirectional long short-term memory and bidirectional encoder representations from transformers (BERT). Findings The authors conducted experiments on both English and multilingual data sets for genre classification. The results show that the BERT model achieved the best accuracy on the English data set, whereas cross-lingual language model pretraining based on RoBERTa (XLM-RoBERTa) performed the best on the multilingual data set. This study found that songs in the metal genre were the most accurately labeled, as their text style and topics were the most distinct from other genres. On the contrary, songs from the pop and rock genres were more challenging to differentiate. This study also compared the impact of different word embeddings on the classification task and found that models with GloVe word embeddings outperformed Word2Vec and the learning embedding layer. Originality/value This study presents the implementation, testing and comparison of various machine and deep learning models for genre classification. The results demonstrate that transformer models, including BERT, robustly optimized BERT pretraining approach, distilled bidirectional encoder representations from transformers, bidirectional and auto-regressive transformers and XLM-RoBERTa, outperformed other models.

The medial temporal lobe supports the quality of visual short-term memory representation.

The quality of short-term memory (STM) underlies our ability to recall the exact details of a recent event, yet how the human brain enables this core cognitive function remains poorly understood. Here we use multiple experimental approaches to test the hypothesis that the quality of STM, such as its precision or fidelity, relies on the medial temporal lobe (MTL), a region commonly associated with the ability to distinguish similar information remembered in long-term memory. First, with intracranial recordings, we find that delay-period MTL activity retains item-specific STM content that is predictive of subsequent recall precision. Second, STM recall precision is associated with an increase in the strength of intrinsic MTL-to-neocortical functional connections during a brief retention interval. Finally, perturbing the MTL through electrical stimulation or surgical removal can selectively reduce STM precision. Collectively, these findings provide converging evidence that the MTL is critically involved in the quality of STM representation.

Neural response generation for task completion using conversational knowledge graph.

Effective dialogue generation for task completion is challenging to build. The task requires the response generation system to generate the responses consistent with intent and slot values, have diversity in response and be able to handle multiple domains. The response also needs to be context relevant with respect to the previous utterances in the conversation. In this paper, we build six different models containing Bi-directional Long Short Term Memory (Bi-LSTM) and Bidirectional Encoder Representations from Transformers (BERT) based encoders. To effectively generate the correct slot values, we implement a copy mechanism at the decoder side. To capture the conversation context and the current state of the conversation we introduce a simple heuristic to build a conversational knowledge graph. Using this novel algorithm we are able to capture important aspects in a conversation. This conversational knowledge-graph is then used by our response generation model to generate more relevant and consistent responses. Using this knowledge-graph we do not need the entire utterance history, rather only the last utterance to capture the conversational context. We conduct experiments showing the effectiveness of the knowledge-graph in capturing the context and generating good response. We compare these results against hierarchical-encoder-decoder models and show that the use of triples from the conversational knowledge-graph is an effective method to capture context and the user requirement. Using this knowledge-graph we show an average performance gain of 0.75 BLEU score across different models. Similar results also hold true across different manual evaluation metrics.

Fake News Encoder Classifier (FNEC) for Online Published News Related to COVID-19 Vaccines

In the past few years, social media and online news platforms have played an essential role in distributing news content rapidly. Consequently. verification of the authenticity of news has become a major challenge. During the COVID-19 outbreak, misinformation and fake news were major sources of confusion and insecurity among the general public. In the first quarter of the year 2020, around 800 people died due to fake news relevant to COVID-19. The major goal of this research was to discover the best learning model for achieving high accuracy and performance. A novel case study of the Fake News Classification using ELECTRA model, which achieved 85.11% accuracy score, is thus reported in this manuscript. In addition to that, a new novel dataset called COVAX-Reality containing COVID-19 vaccine-related news has been contributed. Using the COVAX-Reality dataset, the performance of FNEC is compared to several traditional learning models i.e., Support Vector Machine (SVM), Naive Bayes (NB), Passive Aggressive Classifier (PAC), Long Short-Term Memory (LSTM), Bi-directional LSTM (Bi-LSTM) and Bi-directional Encoder Representations from Transformers (BERT). For the evaluation of FNEC, standard metrics (Precision, Recall, Accuracy, and F1-Score) were utilized.

Self-Supervised Time Series Classification Based on LSTM and Contrastive Transformer

Time series data has attached extensive attention as multi-domain data, but it is difficult to analyze due to its high dimension and few labels. Self-supervised representation learning provides an effective way for processing such data. Considering the frequency domain features of the time series data itself and the contextual feature in the classification task, this paper proposes an unsupervised Long Short-Term Memory (LSTM) and contrastive transformer-based time series representation model using contrastive learning. Firstly, transforming data with frequency domain-based augmentation increases the ability to represent features in the frequency domain. Secondly, the encoder module with three layers of LSTM and convolution maps the augmented data to the latent space and calculates the temporal loss with a contrastive transformer module and contextual loss. Finally, after self-supervised training, the representation vector of the original data can be got from the pre-trained encoder. Our model achieves satisfied performances on Human Activity Recognition (HAR) and sleepEDF real-life datasets.

A text mining-based approach for understanding Chinese railway incidents caused by electromagnetic interference

The high-speed railway is a deeply coupled system with strong and weak electrical equipment, while complex electromagnetic interference (EMI) consequently brings potential risks and hazards to signaling safety. Since the incident reports on signaling failure intrinsically reflect the generation and evolution mechanism of equipment failures, relying on text mining technology, this paper tries to extract failure-related entities and constructs a knowledge graph to clarify the negative impact of the on-site electromagnetic environment. Firstly, based on convolutional neural networks (CNN), a supervised deep learning model for Chinese text classification is established to generate a corpus containing only railway failures caused by EMI. Then, the bidirectional long short-term memory (BiLSTM) and bidirectional encoder representations from transformers (BERT) algorithms are adopted to build the named entity recognition (NER) model. A NER algorithm more suitable for Chinese text features is proposed through ensemble modeling, training verification, and comparative evaluation. Finally, the knowledge storage and visualization of relational graph construction based on the Neo4j database are realized according to the obtained failure-related entities. This knowledge topology network effectively explores the inherent relationship between EMI factors and railway safety, as well as provides support for improving the safety assessment and enhancing the anti-interference performance of the equipment.

Development of Internet suicide message identification and the Monitoring-Tracking-Rescuing model in Taiwan

BackgroundSuicide messages can be transmitted infinitely online; the Internet is influential in suicide prevention. Identifying suicide risks online via artificial technological advances may help predict suicide. MethodsWe built a classifier that detects open messages containing suicidal ideation or behavior-related words in social media via text mining methods and developed the Monitoring-Tracking-Rescuing model, which links data monitoring and tracking to high-risk suicide rescues. Natural language processing (NLP) techniques such as Long Short-Term Memory and Bidirectional Encoder Representations from Transformers were applied to online posts of common social media sites in Taiwan. This model uses a two-step high-risk identification procedure: an automatic prediction process using NLP to classify suicide-risk levels, followed by professional validation by a senior psychiatrist and a nursing faculty specialized in suicidology. ResultsFrom a dataset containing 404 high-risk and 2226 no- or low-risk articles, the sensitivity and specificity of our model reached 80 %. LimitationsThe model is limited to data platforms that can be “crawled” and excludes suicide-risk content from graphics, video and audio files. Additionally, machine learning does not provide the best recognition rate from complex online messages. Keywords for high-risk suicide in long articles are difficult to interpret using this model. Finally, the model lacks keywords for suicide-protective factors. ConclusionsArtificial intelligence techniques may help detect and monitor high-risk suicide posts and inform mental health professionals of these posts. Periodic tracking plus manual validation to determine risk levels are recommended to enhance the reliability and effectiveness of Internet suicide-prevention tasks.

Adult age differences in specific and gist associative episodic memory across short- and long-term retention intervals.

Age-related deficits in associative episodic memory have been widely reported, but recent research suggests that some of these deficits occur for highly specific but not gist representations. It remains undetermined whether older adults' deficits in specific associative episodic memory, observed in long-term memory, are also present in short-term memory. We used a continuous associative recognition task to address this question. Fifty young and 50 older adults studied face-scene pairs, with memory tests occurring in both short-term and long-term memory. Memory tests featured intact (old) pairs, related (similar) pairs, and unrelated (dissimilar) pairs. On short-term memory tests, older adults were less accurate in classifying related pairs, which was manifest by age-related reductions in the probability of retrieving specific memory to engage in recollection rejection. However, older adults were capable of remembering specific details in short-term memory for intact probes but were less likely to remember specific details in long-term memory. Finally, older adults were always as capable as younger adults of remembering gist details. Results suggest that older adults do at least partially encode specific representations in short-term memory, and their access to these specific representations is cue dependent-they can do so when there is a large correspondence between encoding and retrieval conditions but are less likely to engage in deeper elaboration at retrieval. This limits their ability to remember specific details of associations to suppress false recognitions in short-term memory and to engage in veridical recognition in long-term memory. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Active vision in sight recovery individuals with a history of long-lasting congenital blindness.

What we see is intimately linked to how we actively and systematically explore the world through eye movements. However, it is unknown to what degree visual experience during early development is necessary for such systematic visual exploration to emerge. The present study investigated visual exploration behavior in 10 human participants whose sight had been restored only in childhood or adulthood, after a period of congenital blindness because of dense bilateral congenital cataracts. Participants freely explored real-world images while their eye movements were recorded. Despite severe residual visual impairments and gaze instability (nystagmus), visual exploration patterns were preserved in individuals with reversed congenital cataract. Modeling analyses indicated that, similar to healthy control subjects, visual exploration in individuals with reversed congenital cataract was based on the low-level (luminance contrast) and high-level (object components) visual content of the images. Moreover, participants used visual short-term memory representations for narrowing down the exploration space. More systematic visual exploration in individuals with reversed congenital cataract was associated with better object recognition, suggesting that active vision might be a driving force for visual system development and recovery. The present results argue against a sensitive period for the development of neural mechanisms associated with visual exploration.

Comparative research on structure function recognition based on deep learning

PurposeThe research on structure function recognition mainly concentrates on identifying a specific part of academic literature and its applicability in the multidiscipline perspective. A specific part of academic literature, such as sentences, paragraphs and chapter contents are also called a level of academic literature in this paper. There are a few comparative research works on the relationship between models, disciplines and levels in the process of structure function recognition. In view of this, comparative research on structure function recognition based on deep learning has been conducted in this paper.Design/methodology/approachAn experimental corpus, including the academic literature of traditional Chinese medicine, library and information science, computer science, environmental science and phytology, was constructed. Meanwhile, deep learning models such as convolutional neural networks (CNN), long and short-term memory (LSTM) and bidirectional encoder representation from transformers (BERT) were used. The comparative experiments of structure function recognition were conducted with the help of the deep learning models from the multilevel perspective.FindingsThe experimental results showed that (1) the BERT model performed best, with F1 values of 78.02, 89.41 and 94.88%, respectively at the level of sentence, paragraph and chapter content. (2) The deep learning models performed better on the academic literature of traditional Chinese medicine than on other disciplines in most cases, e.g. F1 values of CNN, LSTM and BERT, respectively arrived at 71.14, 69.96 and 78.02% at the level of sentence. (3) The deep learning models performed better at the level of chapter content than other levels, the maximum F1 values of CNN, LSTM and BERT at 91.92, 74.90 and 94.88%, respectively. Furthermore, the confusion matrix of recognition results on the academic literature was introduced to find out the reason for misrecognition.Originality/valueThis paper may inspire other research on structure function recognition, and provide a valuable reference for the analysis of influencing factors.

Identification of hand-foot syndrome from cancer patients' blog posts: BERT-based deep-learning approach to detect potential adverse drug reaction symptoms.

Early detection and management of adverse drug reactions (ADRs) is crucial for improving patients’ quality of life. Hand-foot syndrome (HFS) is one of the most problematic ADRs for cancer patients. Recently, an increasing number of patients post their daily experiences to internet community, for example in blogs, where potential ADR signals not captured through routine clinic visits can be described. Therefore, this study aimed to identify patients with potential ADRs, focusing on HFS, from internet blogs by using natural language processing (NLP) deep-learning methods. From 10,646 blog posts, written in Japanese by cancer patients, 149 HFS-positive sentences were extracted after pre-processing, annotation and scrutiny by a certified oncology pharmacist. The HFS-positive sentences described not only HFS typical expressions like “pain" or “spoon nail”, but also patient-derived unique expressions like onomatopoeic ones. The dataset was divided at a 4 to 1 ratio and used to train and evaluate three NLP deep-learning models: long short-term memory (LSTM), bidirectional LSTM and bidirectional encoder representations from transformers (BERT). The BERT model gave the best performance with precision 0.63, recall 0.82 and f1 score 0.71 in the HFS user identification task. Our results demonstrate that this NLP deep-learning model can successfully identify patients with potential HFS from blog posts, where patients’ real wordings on symptoms or impacts on their daily lives are described. Thus, it should be feasible to utilize patient-generated text data to improve ADR management for individual patients.

Mul-SNO: A Novel Prediction Tool for S-Nitrosylation Sites Based on Deep Learning Methods.

Protein s-nitrosylation (SNO) is one of the most important post-translational modifications and is formed by the covalent modification of nitric oxide and cysteine residues. Extensive studies have shown that SNO plays a pivotal role in the plant immune response and treating various major human diseases. In recent years, SNO sites have become a hot research topic. Traditional biochemical methods for SNO site identification are time-consuming and costly. In this study, we developed an economical and efficient SNO site prediction tool named Mul-SNO. Mul-SNO ensembled current popular and powerful deep learning model bidirectional long short-term memory (BiLSTM) and bidirectional encoder representations from Transformers (BERT). Compared with existing state-of-the-art methods, Mul-SNO obtained better ACC of 0.911 and 0.796 based on 10-fold cross-validation and independent data sets, respectively.

Fast prediction of distances between synthetic routes with deep learning

We expand the recent work on clustering of synthetic routes and train a deep learning model to predict the distances between arbitrary routes. The model is based on a long short-term memory representation of a synthetic route and is trained as a twin network to reproduce the tree edit distance (TED) between two routes. The machine learning approach is approximately two orders of magnitude faster than the TED approach and enables clustering many more routes from a retrosynthesis route prediction. The clusters have a high degree of similarity to the clusters given by the TED-based approach and are accordingly intuitive and explainable. We provide the developed model as open-source.

Joint bidding and pricing for electricity retailers based on multi-task deep reinforcement learning

The single-task deep reinforcement learning (STDRL)-based methods solve the joint bidding and pricing problem for the electricity retailer in a hierarchical electricity market by defining a bidding policy and a pricing policy separately, which may suffer from low learning efficiency, time-consuming training and local optimization. To deal with these issues, this paper proposes a novel Multi-task Deep reinforcement learning approach for Joint Bidding and Pricing (MDJBP) optimization model. MDJBP can deal with the bidding and pricing tasks concurrently through a shared long short-term memory (LSTM) representation network to distill meaningful temporal characteristics from high-dimensional environment states. Furthermore, we develop a deep neural network (DNN) structure consisting a regression branch for bidding task and a soft actor-critic (SAC) branch for pricing task with automating entropy adjustment and adaptive loss weighting to implement MDJBP. The proposed multi-task deep reinforcement learning (MTDRL)-based method is tested with the IEEE 30-bus system. Numerical results show that the proposed methodology succeeds in giving an optimal joint bidding and pricing policy by fully exploiting commonalities and differences between bidding task and pricing task, and thereby boosts the profit, improves learning efficiency, reduces training time, and enhances stability.