Text Mining Strategies Research Articles

Text Mining Strategies: RoBERTa Optimization for Efficient Pain Assessment in Hospice Care

The hospice unit in medical care offers comprehensive, personalized care to patients, yet the recent epidemic and associated illnesses have strained medical resources, leading to a shortage in capacity. The necessity for frequent physiological documentation and patient assessments places a considerable burden on the nursing staff, particularly in the context of limited personnel. This study addresses this challenge by leveraging natural language processing to aid in the evaluation of pain indices, aiming to enhance implementation quality and reduce associated costs. Three BERT models— BERT, MacBERT, and RoBERTa were employed for training purposes. Among these models, RoBERTa demonstrated exceptional performance, achieving an impressive accuracy rate of 99%. This research highlights the potential of natural language processing tools, specifically the RoBERTa model, in alleviating the workload of nursing staff and improving the efficiency of pain assessment in hospice care during times of heightened demand and limited resources.

Artificial intelligence based social robots in the process of student mental health diagnosis

This paper in order to achieve the application of artificial intelligence based social robots in the process of student mental health diagnosis. When designing the architecture of social robots, factors such as interactivity, adaptability, and scalability were taken into consideration to ensure that they possess human like interaction characteristics and flexibility. Subsequently, a model was constructed based on deep learning technology to achieve functions such as sentiment classification, text mining, and optimization strategies. The input data set of the training model comes from the user’s interaction records and behavior data on the Internet social platform, as well as the user’s feedback information in the process of using the robot. The research on psychological data classification has constructed corresponding algorithms based on pointer networks and text models to achieve text feature extraction and classification. The psychological emotion mining module extracts emotional states from user discourse and maps them to corresponding categories of psychological problems. Finally, based on the user input question content, classify and optimize psychological problems. Research has shown that the robot has certain accuracy and practicality in data classification and student mental health diagnosis.

Demand and satisfaction analysis of short health videos among Chinese urban youth: A mixed-methods study based on the KANO model

Short videos have become a powerful medium in health communication. This study explores the information needs and satisfaction of Chinese urban youth regarding short health videos, guided by the KANO model. The research was conducted in four stages: text mining, online survey, demand analysis, and emotional strategy analysis. During the text mining stage, we used GooSeeker software to extract 26,108 health-tagged short video entries from Douyin and identified 21 unique themes through a collinear network analysis. In the online survey stage, we gathered data on urban youth’s preferences for these themes. Using the KANO model and Better-Worse analysis in the demand analysis stage, we categorized health video demands and gained insights into the preferred content. In the emotional strategy analysis stage, we examined how different emotional strategies like appeals to fear and hope influenced content effectiveness. Findings show that content related to health science, tea drinking, popular news, and food safety significantly enhances satisfaction. Conversely, information on refuting rumors, epidemic prevention, and control, as well as authoritative views, tends to lower satisfaction, possibly due to a trust crisis caused by a mismatch between demand and supply. This study suggests that content creators can boost engagement and satisfaction by focusing on preferred themes. It also highlights the varying impacts of information sources and emotional strategies on the health video preferences of Chinese urban youth. The insights from this research provide a foundation for user-centric content creation and platform development in health communication.

Melatonin and cancer: Exploring gene networks and functional categories

While melatonin is known for its multifaceted properties and its potential to combat cancer, there has been limited exploration of the cancer-melatonin interaction at the gene network level. One of the ways to better understand the molecular mechanisms of melatonin’s anti-cancer effects is to use text-mining strategies to extract relevant information that creates knowledge networks of entities and their associations. In this study, we mined gene-publication associations to search for genes most relevant to the terms of “melatonin” and “cancer”. A total of 152 genes were identified and ranked, among which 15 were kinase-related and three G-protein coupled receptor genes. The hub genes (STAT3, JUN, TP53, MAPK3, EP300, SRC, HSP90AA1, AKT1, ESR1, and IL6) were involved with several pathways in cancer. After examining the melatonin-treated cancers, we mapped 25 upregulated and 51 downregulated genes; these were strongly associated with cancer hallmarks such as resisting cell death, sustaining proliferative signaling, and inducing invasion and metastasis. Upregulated genes showed molecular functions including apoptotic protease activator, caspase activator, enzyme regulator, and protein binding, whereas the downregulated genes affected protein kinase activities, transcription factor binding, protein, enzyme, DNA, and promoter bindings. By connecting gene subsets, we detected a closer relationship among breast, hepatocellular, prostate, and oral cancers, in addition to neuroblastoma and osteosarcoma in terms of changes in melatonin-related signaling pathways. TCGA data were analyzed to understand the impact of gene signatures on survival of patients, and melatonin-downregulated genes were associated with longer survival of patients with glioblastoma, bladder, breast, colon, stomach, liver, lung, and ovarian carcinomas. These results provide a global view of gene interaction networks in melatonin-treated cancers and their functional value, opening new opportunities to consider melatonin for cancer therapy.

Novel biomarker genes for the prediction of post-hepatectomy survival of patients with NAFLD-related hepatocellular carcinoma

BackgroundThe incidence and prevalence of nonalcoholic fatty liver disease related hepatocellular carcinoma (NAFLD-HCC) are rapidly increasing worldwide. This study aimed to identify biomarker genes for prognostic prediction model of NAFLD-HCC hepatectomy by integrating text-mining, clinical follow-up information, transcriptomic data and experimental validation.MethodsThe tumor and adjacent normal liver samples collected from 13 NAFLD-HCC and 12 HBV-HCC patients were sequenced using RNA-Seq. A novel text-mining strategy, explainable gene ontology fingerprint approach, was utilized to screen NAFLD-HCC featured gene sets and cell types, and the results were validated through a series of lab experiments. A risk score calculated by the multivariate Cox regression model using discovered key genes was established and evaluated based on 47 patients’ follow-up information.ResultsDifferentially expressed genes associated with NAFLD-HCC specific tumor microenvironment were screened, of which FABP4 and VWF were featured by previous reports. A risk prediction model consisting of FABP4, VWF, gender and TNM stage were then established based on 47 samples. The model showed that overall survival in the high-risk score group was lower compared with that in the low-risk score group (p = 0.0095).ConclusionsThis study provided the landscape of NAFLD-HCC transcriptome, and elucidated that our model could predict hepatectomy prognosis with high accuracy.

An Atlas of the Knee Joint Proteins and Their Role in Osteoarthritis Defined by Literature Mining

Osteoarthritis (OA) is the most prevalent rheumatic pathology. However, OA is not simply a process of wear and tear affecting articular cartilage but rather a disease of the entire joint. One of the most common locations of OA is the knee. Knee tissues have been studied using molecular strategies, generating a large amount of complex data. As one of the goals of the Rheumatic and Autoimmune Diseases initiative of the Human Proteome Project, we applied a text-mining strategy to publicly available literature to collect relevant information and generate a systematically organized overview of the proteins most closely related to the different knee components. To this end, the PubPular literature-mining software was employed to identify protein-topic relationships and extract the most frequently cited proteins associated with the different knee joint components and OA. The text-mining approach searched over eight million articles in PubMed up to November 2022. Proteins associated with the six most representative knee components (articular cartilage, subchondral bone, synovial membrane, synovial fluid, meniscus, and cruciate ligament) were retrieved and ranked by their relevance to the tissue and OA. Gene ontology analyses showed the biological functions of these proteins. This study provided a systematic and prioritized description of knee-component proteins most frequently cited as associated with OA. The study also explored the relationship of these proteins to OA and identified the processes most relevant to proper knee function and OA pathophysiology.

A Brief Survey of Text Document Classification Algorithms and Processes

The exponential growth of unstructured data is one of the most critical challenges in data mining, text analytics, or data analytics. Around 80% of the world's data are available in unstructured format and most are left unattended due to the complexity of its analysis. It is a great challenge to guarantee the quality of the text document classifier that classifies documents based on user preferences because of large-scale terms and data patterns. The World Wide Web is growing rapidly and the availability of electronic documents is also increasing. Therefore, the automatic categorization of documents is the key factor for the systematic organization of information and knowledge discovery. Most existing widespread text mining and classification strategies have adopted term-based approaches. However, the problems of polysemy and synonymy in such approaches are of great concern. To classify documents based on their context, the context-based approach is needed to be followed. Semantic analysis of the text overcomes the limitations of the term-based approach and it also enhances the accuracy of the classifiers. This paper aims to highlight the important algorithms, techniques, and methodologies that can be used for text document classification. Furthermore, the paper also provides a review of the different stages of Text Document Classification.

Text and Social Analytics with Fusion Techniques Enhance Hospital Health Management

the impact of social analytics on hospital health management: a multilevel fusion approach for data-driven decision-making and brand improvement. The hospital health management center should use feature extraction techniques to learn more about customers' feelings towards their services and optimize their business strategies and promotions accordingly. The proposed multi-levelhybrid level fusion system architectures can effectively integrate dataimages from multiple sources, including social networks, to collect and process essential data for score level and rank level decision-making. This approach leverages intelligent techniques, such as deep learning models, fuzzy logic, and optimization algorithms, to improve fusion scores and achieve optimal fusion performance. The proposed framework can also be extended to various applications, including multimedia data fusion, e-systems data fusion, and spatial data fusion, to enable intelligent systems for information fusion and decision-making in diverse domains. Therefore, this paper proposes Improved Customer Relation and Business Operations (ICR-BO) to enhance customer relationships in business development using text and social analytics. A case study is carried out to explore the online debate of computer brands operated in hospital environments and Twitter suppliers. The authors used text-mining strategies and social analytics to analyze business operations. Social Media uses data sets to view important observations and trends to identify consumer awareness after collecting critical tweets using Twitter search. The experimental results show that ICR-BO achieves the highest customer relation compared to other existing methods.

Genomic and proteomic biomarker landscape in clinical trials

The use of molecular biomarkers to support disease diagnosis, monitor its progression, and guide drug treatment has gained traction in the last decades. While only a dozen biomarkers have been approved for their exploitation in the clinic by the FDA, many more are evaluated in the context of translational research and clinical trials. Furthermore, the information on which biomarkers are measured, for which purpose, and in relation to which conditions are not readily accessible: biomarkers used in clinical studies available through resources such as ClinicalTrials.gov are described as free text, posing significant challenges in finding, analyzing, and processing them by both humans and machines. We present a text mining strategy to identify proteomic and genomic biomarkers used in clinical trials and classify them according to the methodologies by which they are measured. We find more than 3000 biomarkers used in the context of 2600 diseases. By analyzing this dataset, we uncover patterns of use of biomarkers across therapeutic areas over time, including the biomarker type and their specificity. These data are made available at the Clinical Biomarker App at https://www.disgenet.org/biomarkers/, a new portal that enables the exploration of biomarkers extracted from the clinical studies available at ClinicalTrials.gov and enriched with information from the scientific literature. The App features several metrics that assess the specificity of the biomarkers, facilitating their selection and prioritization. Overall, the Clinical Biomarker App is a valuable and timely resource about clinical biomarkers, to accelerate biomarker discovery, development, and application.

The Immunometabolic Atlas: A tool for design and interpretation of metabolomics studies in immunology.

Immunometabolism, which concerns the interplay between metabolism and the immune system, is increasingly recognized as a potential source of novel drug targets and biomarkers. In this context, the use of metabolomics to identify metabolic characteristics associated with specific functional immune response processes is of value. Currently, there is a lack of tools to determine known associations between metabolites and immune processes. Consequently, interpretation of metabolites in metabolomics studies in terms of their role in the immune system, or selection of the most relevant metabolite classes to include in metabolomics studies, is challenging. Here, we describe the Immunometabolic Atlas (IMA), a public web application and library of R functions to infer immune processes associated with specific metabolites and vice versa. The IMA derives metabolite-immune process associations utilizing a protein-metabolite network analysis algorithm that associates immune system-associated annotated proteins in Gene Ontology to metabolites. We evaluated IMA inferred metabolite-immune system associations using a text mining strategy, identifying substantial overlap, but also demonstrating a significant chemical space of immune system-associated metabolites that should be confirmed experimentally. Overall, the IMA facilitates the interpretation and design of immunometabolomics studies by the association of metabolites to specific immune processes.

The Immunometabolic Atlas: A tool for design and interpretation of metabolomics studies in immunology

Immunometabolism, which concerns the interplay between metabolism and the immune system, is increasingly recognized as a potential source of novel drug targets and biomarkers. In this context, the use of metabolomics to identify metabolic characteristics associated with specific functional immune response processes is of value. Currently, there is a lack of tools to determine known associations between metabolites and immune processes. Consequently, interpretation of metabolites in metabolomics studies in terms of their role in the immune system, or selection of the most relevant metabolite classes to include in metabolomics studies, is challenging. Here, we describe the Immunometabolic Atlas (IMA), a public web application and library of R functions to infer immune processes associated with specific metabolites and vice versa. The IMA derives metabolite-immune process associations utilizing a protein-metabolite network analysis algorithm that associates immune system-associated annotated proteins in Gene Ontology to metabolites. We evaluated IMA inferred metabolite-immune system associations using a text mining strategy, identifying substantial overlap, but also demonstrating a significant chemical space of immune system-associated metabolites that should be confirmed experimentally. Overall, the IMA facilitates the interpretation and design of immunometabolomics studies by the association of metabolites to specific immune processes.

ExTRI: Extraction of transcription regulation interactions from literature

The regulation of gene transcription by transcription factors is a fundamental biological process, yet the relations between transcription factors (TF) and their target genes (TG) are still only sparsely covered in databases. Text-mining tools can offer broad and complementary solutions to help locate and extract mentions of these biological relationships in articles. We have generated ExTRI, a knowledge graph of TF-TG relationships, by applying a high recall text-mining pipeline to MedLine abstracts identifying over 100,000 candidate sentences with TF-TG relations. Validation procedures indicated that about half of the candidate sentences contain true TF-TG relationships. Post-processing identified 53,000 high confidence sentences containing TF-TG relationships, with a cross-validation F1-score close to 75%. The resulting collection of TF-TG relationships covers 80% of the relations annotated in existing databases. It adds 11,000 other potential interactions, including relationships for ~100 TFs currently not in public TF-TG relation databases. The high confidence abstract sentences contribute 25,000 literature references not available from other resources and offer a wealth of direct pointers to functional aspects of the TF-TG interactions. Our compiled resource encompassing ExTRI together with publicly available resources delivers literature-derived TF-TG interactions for more than 900 of the 1500–1600 proteins considered to function as specific DNA binding TFs. The obtained result can be used by curators, for network analysis and modelling, for causal reasoning or knowledge graph mining approaches, or serve to benchmark text mining strategies.

Mapping the Sensory Fingerprint of Swedish Beer Market through Text and Data Mining and Multivariate Strategies

The continuous increase of online data with consumers’ and experts’ reviews and preferences is a potential tool for sensory characterization. The present work aims to overview the Swedish beer market and understand the sensory fingerprint of Swedish beers based on text data extracted from the Swedish alcohol retail monopoly (Systembolaget) website. Different multivariate strategies such as heatmaps, correspondence analysis and hierarchical cluster analysis were used to understand the sensory space of the different beer styles. Additionally, sensory space for specific hop cultivars was also investigated. Results highlighted Gothenburg as the main producing area in Sweden. The style Indian Pale Ale (IPA) is the largest available at the retail monopoly. From a sensory perspective, commonalities and differences were found between beer types and styles. Based on the aroma description, different types of ale and lager can cluster together (such as Porter and Stout and Dark lagers). Additionally, an associative relationship between specific aromas and hop cultivars from text data information was successfully achieved.

A Novel Approach for Detection of Fake News using Long Short Term Memory (LSTM)

Online media for news consumption has doubtful advantages. From one perspective, it has minimal expense, simple access, and fast dispersal of data which leads individuals to search out and devour news from online media. On the other hand, it increases the wide spread of "counterfeit news", i.e., inferior quality news with purposefully bogus data. The broad spread of fake news contrarily affects people and society. Hence, fake news detection in social media has become an emerging research topic that is drawing attention from various researchers. In past, many creators proposed the utilization of text mining procedures and AI strategies to examine textual data and helps to foresee the believability of news. With more computational capacities and to deal with enormous datasets, deep learning models present a better presentation over customary text mining strategies and AI methods. Normally deep learning model, for example, LSTM model can identify complex patterns in the data. Long short term memory is a tree organized recurrent neural network (RNN) used to examine variable length sequential information. In our proposed framework we set up a fake news identification model dependent on LSTM neural network. Openly accessible unstructured news datasets are utilized to evaluate the exhibition of the model. The outcome shows the prevalence and exactness of LSTM model over the customary techniques specifically CNN for fake news recognition.

Linking common human diseases to their phenotypes; development of a resource for human phenomics

BackgroundIn recent years a large volume of clinical genomics data has become available due to rapid advances in sequencing technologies. Efficient exploitation of this genomics data requires linkage to patient phenotype profiles. Current resources providing disease-phenotype associations are not comprehensive, and they often do not have broad coverage of the disease terminologies, particularly ICD-10, which is still the primary terminology used in clinical settings.MethodsWe developed two approaches to gather disease-phenotype associations. First, we used a text mining method that utilizes semantic relations in phenotype ontologies, and applies statistical methods to extract associations between diseases in ICD-10 and phenotype ontology classes from the literature. Second, we developed a semi-automatic way to collect ICD-10–phenotype associations from existing resources containing known relationships.ResultsWe generated four datasets. Two of them are independent datasets linking diseases to their phenotypes based on text mining and semi-automatic strategies. The remaining two datasets are generated from these datasets and cover a subset of ICD-10 classes of common diseases contained in UK Biobank. We extensively validated our text mined and semi-automatically curated datasets by: comparing them against an expert-curated validation dataset containing disease–phenotype associations, measuring their similarity to disease–phenotype associations found in public databases, and assessing how well they could be used to recover gene–disease associations using phenotype similarity.ConclusionWe find that our text mining method can produce phenotype annotations of diseases that are correct but often too general to have significant information content, or too specific to accurately reflect the typical manifestations of the sporadic disease. On the other hand, the datasets generated from integrating multiple knowledgebases are more complete (i.e., cover more of the required phenotype annotations for a given disease). We make all data freely available at https://doi.org/10.5281/zenodo.4726713.

The power of engagement: Understanding active social media engagement and the impact on sales in the hospitality industry

This study seeks to understand the most important features of active social media engagement among followers of a food and beverage sales organization, and the impact of social media engagement rate on sales of new products. The study provides empirical evidence that highly engaged social media posts drive firm performance through increased sales. Additionally, the study provides a formula for success by using text mining strategies to uncover most effective use of the social media channels. The text mining methods reveal the best date/time, type of posting, hashtag to use and social media channel that provides the highest engagement level.

Do expectations towards Thai hospitality differ? The views of English vs Chinese speaking travelers

Purpose This paper aims to shed light on differences in the patterns of online rating behavior that Chinese- and English-speaking travelers adopt when making hotel reviews on TripAdvisor. Design/methodology/approach A dual analysis was conducted on 800 online reviews for eight hotel brands. The brands, which are currently operating in Bangkok City, are considered to be representative of their categories. The hotels were selected based on their abilities to meet the quantitative and qualitative requirements of the text mining strategy adopted in this study. Findings The results reveal that, with respect to all of the hotel service attributes (i.e. Service, Cleanliness, Room, Sleep Quality, Location, Value and Overall), the ratings offered by the English-speaking guests were higher than the Chinese-speaking travelers. Based on the qualitative analysis, the ten service attributes which had the most impact on traveler satisfaction were distinguished. These attributes were then classified into three broadly themed categories including intangible service, tangible service and staying experience. The results from the word frequency analysis also helps to clarify which features are drawing attention from travelers from different backgrounds. Originality/value This study contributes to tourism and hospitality literature by confirming certain differences in the online rating behavior of Chinese- and English-speaking hotel guests.

From cesspool to fortified paradise: analyzing news media territorial assemblages of rehabilitating Boracay Island, Western Philippines

Sustainability initiatives have expanded beyond the rehabilitation of physical environments towards the reconstruction of discourses that shape human-nature interactions. In the present study, the discursive-materiality of territorial assemblages of sustainability were examined by analyzing tabloid news report coverage of the closure, rehabilitation, and reopening of Boracay, an island in the Western Philippines hailed as one of the most popular beach destinations in Asia. News reports about Boracay from eight local tabloids (n = 507) were collected and analyzed in two phases. The data corpus underwent basic descriptive coding through a text mining strategy called Latent Dirichlet Allocation (LDA), a confirmatory form of analysis that grouped data into clusters of meaning. From these initial clusters, discursive-material assemblages and their respective elements were critically mapped and identified through further interpretative analysis. Throughout the process of rehabilitation, Boracay was (re)constructed from a cesspool into a fortified paradise. These discursive-material assemblages of sustainability are further discussed in relation to the broader political climate of the current administration, as well as, the reproduction of claiming territories of public good in online space.

Sentiment analysis and text categorization of cancer medical records with LSTM

Cancer is one among leading diseases, which affects millions of people and families around the world. Monitoring the mood of such cancer affected people plays a vital part in their treatment. In recent days, social media provides a platform for many people to share their experiences about the cancer through various blogs and communities. In this study, we intended to analyse moods of various cancer affected patients by collecting tweets from different online cancer supported communities. We employed several text mining and machine learning strategies to perform sentiment analysis on a distributed framework and developed a model for easier and faster analysis. The proposed distributed framework with long short-term memory (LSTM) neural network is an alternative to the conventional sentiment analysis approaches in analysing large volumes of data in a potential flow. The effectiveness of proposed framework was evaluated on the proposed dataset (corpus-1) and other two benchmark datasets like Health news Tweets (corpus-2) and Medical abstracts (corpus-3). The performance of each text mining and classification method was separately evaluated on three datasets and compared to each other. The results proved that the proposed approach performed better among the other methods in terms of both accuracy and execution time.

Semi-automated text mining strategies for identifying rare causes of injuries from emergency room triage data

Human coders, in many organizations conducting injury surveillance, routinely assign External-cause-of-injury codes (E-codes) to short narratives describing the incident, transcribed by triage nurses or others in hospital emergency rooms or other settings. Machine learning (ML) models trained on coded injury narratives can accurately assign E-codes to a large portion of the data, but tend to poorly predict cases falling into rare categories. In this study, we examined several ways of filtering out cases for human review that were likely to belong to rare categories from the predictions of Logistic Regression and Naïve Bayes classifiers for a manually-coded emergency department triage dataset of approximately 500,000 cases, collected between years 2002–2012, provided by the Queensland Injury Surveillance Unit. The ML models were trained using 90% of the data and the filtering approaches were evaluated on a prediction set comprised of the remaining cases. Cost analysis was also performed to compare the efficiency of each filtering method. The results showed that each filtering method greatly improved the ability to detect rare categories. Filtering using expert-designed causal linguistic rules combined with Logistic Regression prediction strength was found to be the most efficient approach. Several completely automated filtering approaches were also found to be effective.