Filtering big data from social media--Building an early warning system for adverse drug reactions

OBJECTIVES

Adverse drug reactions (ADRs) are believed to be a leading cause of death in the world. Pharmacovigilance systems are aimed at early detection of ADRs. With the popularity of social media, Web forums and discussion boards become important sources of data for consumers to share their drug use experience, as a result may provide useful information on drugs and their adverse reactions. In this study, we propose an automated ADR related posts filtering mechanism using text classification methods. In real-life settings, ADR related messages are highly distributed in social media, while non-ADR related messages are unspecific and topically diverse. It is expensive to manually label a large amount of ADR related messages (positive examples) and non-ADR related messages (negative examples) to train classification systems. To mitigate this challenge, we examine the use of a partially supervised learning classification method to automate the process.

METHODS

We propose a novel pharmacovigilance system leveraging a Latent Dirichlet Allocation modeling module and a partially supervised classification approach. We select drugs with more than 500 threads of discussion, and collect all the original posts and comments of these drugs using an automatic Web spidering program as the text corpus. Various classifiers were trained by varying the number of positive examples and the number of topics. The trained classifiers were applied to 3000 posts published over 60 days. Top-ranked posts from each classifier were pooled and the resulting set of 300 posts was reviewed by a domain expert to evaluate the classifiers.

RESULTS

Compare to the alternative approaches using supervised learning methods and three general purpose partially supervised learning methods, our approach performs significantly better in terms of precision, recall, and the F measure (the harmonic mean of precision and recall), based on a computational experiment using online discussion threads from Medhelp.

CONCLUSIONS

Our design provides satisfactory performance in identifying ADR related posts for post-marketing drug surveillance. The overall design of our system also points out a potentially fruitful direction for building other early warning systems that need to filter big data from social media networks.

Modeling the public attitude towards organic foods: a big data and text mining approach

This study aims to identify the topics that users post on Twitter about organic foods and to analyze the emotion-based sentiment of those tweets. The study addresses a call for an application of big data and text mining in different fields of research, as well as proposes more objective research methods in studies on food consumption. There is a growing interest in understanding consumer choices for foods which are caused by the predominant contribution of the food industry to climate change. So far, customer attitudes towards organic food have been studied mostly with self-reported methods, such as questionnaires and interviews, which have many limitations. Therefore, in the present study, we used big data and text mining techniques as more objective methods to analyze the public attitude about organic foods. A total of 43,724 Twitter posts were extracted with streaming Application Programming Interface (API). Latent Dirichlet Allocation (LDA) algorithm was applied for topic modeling. A test of topic significance was performed to evaluate the quality of the topics. Public sentiment was analyzed based on the NRC emotion lexicon by utilizing Syuzhet package. Topic modeling results showed that people discuss on variety of themes related to organic foods such as plant-based diet, saving the planet, organic farming and standardization, authenticity, and food delivery, etc. Sentiment analysis results suggest that people view organic foods positively, though there are also people who are skeptical about the claims that organic foods are natural and free from chemicals and pesticides. The study contributes to the field of consumer behavior by implementing research methods grounded in text mining and big data. The study contributes also to the advancement of research in the field of sustainable food consumption by providing a fresh perspective on public attitude toward organic foods, filling the gaps in existing literature and research.

KEAHT: A Knowledge-Enriched Attention-Based Hybrid Transformer Model for Social Sentiment Analysis

Social media materialized as an influential platform that allows people to share their views on global and local issues. Sentiment analysis can handle these massive amounts of unstructured reviews and convert them into meaningful opinions. Undoubtedly, COVID-19 originated as the enormous challenge across the world that physically and financially bruted humankind. Meanwhile, farmers' protests shook up the world against three pieces of legislation passed by the Indian government. Hence, an artificial intelligence-based sentiment model is needed for suggesting the right direction toward outbreaks. Although Deep Neural Network (DNN) gained popularity in sentiment analysis applications, these still have a limitation of sequential training, high-dimension feature space, and equal feature importance distribution. In addition, inaccurate polarity scoring and utility-based topic modeling are other challenging aspects of sentiment analysis. It motivates us to propose a Knowledge-Enriched Attention-based Hybrid Transformer (KEAHT) model by enriching the explicit knowledge of Latent Dirichlet Allocation (LDA) topic modeling and lexicalized domain ontology. A pre-trained Bidirectional Encoder Representation from Transformer (BERT) is employed to train within a minimum training corpus. It provides the facility of attention mechanism and can solve complex text problems accurately. A comparative study with existing baselines and recent hybrid models affirms the credibility of the proposed KEAHT in the field of Natural Language Processing (NLP). This model emphasizes artificial intelligence's role in handling the situation of the global pandemic and democratic dispute in a country. Furthermore, two benchmark datasets, namely "COVID-19-Vaccine-Labelled-Tweets" and "Indian-Farmer-Protest-Labelled-Tweets", are also constructed to accommodate future researchers for outlining the essential facts associated with the outbreaks.

Obliviate! Reviewing Neural Fundamentals of Intentional Forgetting from a Meta-Analytic Perspective

Intentional forgetting (IF) is an important adaptive mechanism necessary for correct memory functioning, optimal psychological wellbeing, and appropriate daily performance. Due to its complexity, the neuropsychological processes that give birth to successful intentional forgetting are not yet clearly known. In this study, we used two different meta-analytic algorithms, Activation Likelihood Estimation (ALE) & Latent Dirichlet Allocation (LDA) to quantitatively assess the neural correlates of IF and to evaluate the degree of compatibility between the proposed neurobiological models and the existing brain imaging data. We found that IF involves the interaction of two networks, the main “core regions” consisting of a primarily right-lateralized frontal-parietal circuit that is activated irrespective of the paradigm used and sample characteristics and a second less constrained “supportive network” that involves frontal-hippocampal interactions when IF takes place. Additionally, our results support the validity of the inhibitory or thought suppression hypothesis. The presence of a neural signature of IF that is stable regardless of experimental paradigms is a promising finding that may open new venues for the development of effective clinical interventions.

Depression and Anxiety in Patients with Psoriasis: A Comprehensive Analysis Combining Bibliometrics, Latent Dirichlet Allocation, and HJ-Biplot

Background: Patients with psoriasis often experience psychiatric comorbidities, such as depression and anxiety. These comorbidities can lead to poorer adherence to treatment regimens, reduced effectiveness of therapies, and a heightened disease burden. This study aims to explore the scientific output related to psoriasis, depression, and anxiety using a comprehensive analysis combining bibliometric statistical methods. Methods: The study performed a bibliometric analysis of publications related to psoriasis, depression, and anxiety between 1974 and December 2023. This study employed the Latent Dirichlet Allocation (LDA) algorithm to identify key research topics and used the HJ-Biplot technique to visualize the relationships between publications and research indicators. The inclusion criteria were limited to English-language research articles. Results: Over 49 years, the analysis identified 5059 documents published across 1151 sources. The annual growth rate for research was 12.26%. The Journal of the European Academy of Dermatology and Venereology and The British Journal of Dermatology were found to be the leading journals in this field. The United States emerged as the top contributor, followed by China, Italy, and Germany. The most prevalent research topics were inflammation and cellular function, with a significant focus on patient treatment and the impact of depression and anxiety. Conclusions: This bibliometric analysis underscores the increasing of studies on the comorbidities of depression and anxiety in patients with psoriasis. This study provides a comprehensive overview of research trends and emerging topics in this field, offering valuable insights for future investigations.

Implications of a Twitter data-centred methodology for assessing commuters' perceptions of the Delhi metro in India

Owing to the onset of the new media age, the idea of e-public participation has proven to be a great complement to the limitations of the conventional public participation approach. In this respect, location-based social networks (LBSN) data can prove to be a game shift in this digital era to offer an insight into the commuter perception of service delivery. The paper aims to investigate the potential of using Twitter data to assess commuters’ perceptions of the Delhi metro, India, by presenting a comprehensive methodology for extracting, processing, and interpreting the data. The study extracts Twitter data from the official handle of the Delhi metro, performs semantic and sentiment analysis to comprehend commuters’ concerns and assesses commuters’ sentiments on the predicted concerns. The paper outlines that the current depth of Twitter data is more inclined to instantaneous responses to grievances encountered. Moreover, the analysis presents that for the data extraction period, the topics ‘Ride Safety’ and ‘Crowding’ have the lowest scores, while ‘Personnel Attitude’ and ‘Customer Interface’ have the highest scores. Further, the paper highlights insights gleaned from Twitter data in addition to the aspects included in the conventional satisfaction survey. The paper concludes by outlining the opportunities and limitations of LBSN analytics for effective public transportation decision-making in India.

Exploring Trends in Environmental, Social, and Governance Themes and Their Sentimental Value Over Time

Environmental, social, and governance (ESG) is an indicator that measures a company's non-financial performance. Many firms have recently emphasized the importance of ESG. Ascertaining what topics are being discussed around ESG and how they change over time will contribute significantly to gaining insight into ESG. Using 73,397,870 text data scraped and refined from publicly available Twitter data, this study applied Latent Dirichlet Allocation (LDA) and the dynamic topic model (DTM) to ascertain the hidden structure of the ESG-related document collection and the topics being discussed. The study further conducts a sentiment analysis to examine the sentiment of the general public regarding ESG. Topic modeling shows that various topics regarding ESG are being discussed and evolve over time. Sentiment analysis shows that many people have neutral or positive sentiments toward ESG-related issues. This study contributes to exploring insights into ESG among the public and understanding public reactions toward ESG. We further conclude the study with a discussion of managerial implications and potential future research.

Computational vs. qualitative: analyzing different approaches in identifying networked frames during the Covid-19 crisis

Despite the increasing adaption of automated text analysis in communication studies, its strengths and weaknesses in framing analysis are so far unknown. Fewer efforts have been made to automatic detection of networked frames. Drawing on the recent developments in this field, we harness a comparative exploration, using Latent Dirichlet Allocation (LDA) and a human-driven qualitative coding process on three different samples. Samples were comprised of a dataset of 4,165,177 million tweets collected from Iranian Twittersphere during the Coronavirus crisis, from 21 January, 2020 to 29 April, 2020. Findings showed that while LDA is reliable in identifying the most prominent networked frames, it misses to detects less dominant frames. Our investigation also confirmed that LDA works better on larger datasets and lexical semantics. Finally, we argued that LDA could give us some primary intuitions, but qualitative interpretations are indispensable for understanding the deeper layers of meaning.

Intersections of Big Data and IoT in Academic Publications: A Topic Modeling Approach

As vast amounts of data are generated from various sources such as social media, sensors and online transactions, the analysis of Big Data offers organizations the ability to derive insights and make informed decisions. Simultaneously, IoT connects physical devices, enabling real-time data collection and exchange that transforms interactions within smart homes, cities and industries. The intersection of these fields is essential, leading to innovations such as predictive maintenance, real-time traffic management and personalized solutions. Utilizing a dataset of 8159 publications sourced from the Web of Science database, our research employs Natural Language Processing (NLP) techniques and selective human validation to analyze abstracts, titles, keywords and other useful information, uncovering key themes and trends in both Big Data and IoT research. Six topics are extracted using Latent Dirichlet Allocation. In Topic 1, words like "system" and "energy" are among the most frequent, signaling that Topic 1 revolves around data systems and IoT technologies, likely in the context of smart systems and energy-related applications. Topic 2 focuses on the application of technologies, as indicated by terms such as "technologies", "industry" and "research". It deals with how IoT and related technologies are transforming various industries. Topic 3 emphasizes terms like learning and research, indicating a focus on machine learning and IoT applications. It is oriented toward research involving new methods and models in the IoT domain related to learning algorithms. Topic 4 highlights terms such as smart, suggesting a focus on smart technologies and systems. Topic 5 touches upon the role of digital chains and supply systems, suggesting an industrial focus on digital transformation. Topic 6 focuses on technical aspects such as modeling, system performance and prediction algorithms. It delves into the efficiency of IoT networks with terms like "accuracy", "power" and "performance" standing out.