Dataset Information

Assessing Suicide Risk and Emotional Distress in Chinese Social Media: A Text Mining and Machine Learning Study.

ABSTRACT: Early identification and intervention are imperative for suicide prevention. However, at-risk people often neither seek help nor take professional assessment. A tool to automatically assess their risk levels in natural settings can increase the opportunity for early intervention.The aim of this study was to explore whether computerized language analysis methods can be utilized to assess one's suicide risk and emotional distress in Chinese social media.A Web-based survey of Chinese social media (ie, Weibo) users was conducted to measure their suicide risk factors including suicide probability, Weibo suicide communication (WSC), depression, anxiety, and stress levels. Participants' Weibo posts published in the public domain were also downloaded with their consent. The Weibo posts were parsed and fitted into Simplified Chinese-Linguistic Inquiry and Word Count (SC-LIWC) categories. The associations between SC-LIWC features and the 5 suicide risk factors were examined by logistic regression. Furthermore, the support vector machine (SVM) model was applied based on the language features to automatically classify whether a Weibo user exhibited any of the 5 risk factors.A total of 974 Weibo users participated in the survey. Those with high suicide probability were marked by a higher usage of pronoun (odds ratio, OR=1.18, P=.001), prepend words (OR=1.49, P=.02), multifunction words (OR=1.12, P=.04), a lower usage of verb (OR=0.78, P<.001), and a greater total word count (OR=1.007, P=.008). Second-person plural was positively associated with severe depression (OR=8.36, P=.01) and stress (OR=11, P=.005), whereas work-related words were negatively associated with WSC (OR=0.71, P=.008), severe depression (OR=0.56, P=.005), and anxiety (OR=0.77, P=.02). Inconsistently, third-person plural was found to be negatively associated with WSC (OR=0.02, P=.047) but positively with severe stress (OR=41.3, P=.04). Achievement-related words were positively associated with depression (OR=1.68, P=.003), whereas health- (OR=2.36, P=.004) and death-related (OR=2.60, P=.01) words positively associated with stress. The machine classifiers did not achieve satisfying performance in the full sample set but could classify high suicide probability (area under the curve, AUC=0.61, P=.04) and severe anxiety (AUC=0.75, P<.001) among those who have exhibited WSC.SC-LIWC is useful to examine language markers of suicide risk and emotional distress in Chinese social media and can identify characteristics different from previous findings in the English literature. Some findings are leading to new hypotheses for future verification. Machine classifiers based on SC-LIWC features are promising but still require further optimization for application in real life.

SUBMITTER: Cheng Q

PROVIDER: S-EPMC5525005 | biostudies-other | 2017 Jul

REPOSITORIES: biostudies-other

ACCESS DATA

Similar Datasets

Project description:BackgroundText-based digital media platforms have revolutionized communication and information sharing, providing valuable access to knowledge and understanding in the fields of mental health and suicide prevention.ObjectiveThis systematic review aimed to determine how machine learning and data analysis can be applied to text-based digital media data to understand mental health and aid suicide prevention.MethodsA systematic review of research papers from the following major electronic databases was conducted: Web of Science, MEDLINE, Embase (via MEDLINE), and PsycINFO (via MEDLINE). The database search was supplemented by a hand search using Google Scholar.ResultsOverall, 19 studies were included, with five major themes as to how data analysis and machine learning techniques could be applied: (1) as predictors of personal mental health, (2) to understand how personal mental health and suicidal behavior are communicated, (3) to detect mental disorders and suicidal risk, (4) to identify help seeking for mental health difficulties, and (5) to determine the efficacy of interventions to support mental well-being.ConclusionsOur findings show that data analysis and machine learning can be used to gain valuable insights, such as the following: web-based conversations relating to depression vary among different ethnic groups, teenagers engage in a web-based conversation about suicide more often than adults, and people seeking support in web-based mental health communities feel better after receiving online support. Digital tools and mental health apps are being used successfully to manage mental health, particularly through the COVID-19 epidemic, during which analysis has revealed that there was increased anxiety and depression, and web-based communities played a part in reducing isolation during the pandemic. Predictive analytics were also shown to have potential, and virtual reality shows promising results in the delivery of preventive or curative care. Future research efforts could center on optimizing algorithms to enhance the potential of text-based digital media analysis in mental health and suicide prevention. In addressing depression, a crucial step involves identifying the factors that contribute to happiness and using machine learning to forecast these sources of happiness. This could extend to understanding how various activities result in improved happiness across different socioeconomic groups. Using insights gathered from such data analysis and machine learning, there is an opportunity to craft digital interventions, such as chatbots, designed to provide support and address mental health challenges and suicide prevention.

Project description:ObjectivesThe study aimed to conduct a bibliometric analysis of publications concerning lumbar spondylolisthesis, as well as summarize its research topics and hotspot trends with machine-learning based text mining.MethodsThe data were extracted from the Web of Science Core Collection (WoSCC) database and then analyzed in Rstudio1.3.1 and CiteSpace5.8. Annual publication production and the top-20 productive authors over time were obtained. Additionally, top-20 productive journals and top-20 influential journals were compared by spine-subspecialty or not. Similarly, top-20 productive countries/regions and top-20 influential countries/regions were compared by they were developed countries/regions or not. The collaborative relationship among countries and institutions were presented. The main topics of lumbar spondylolisthesis were classified by Latent Dirichlet allocation (LDA) analysis, and the hotspot trends were indicated by keywords with strongest citation bursts.ResultsUp to 2021, a total number of 4,245 articles concerning lumbar spondylolisthesis were finally included for bibliometric analysis. Spine-subspecialty journals were found to be dominant in the productivity and the impact of the field, and SPINE, EUROPEAN SPINE JOURNAL and JOURNAL OF NEUROSURGERY-SPINE were the top-3 productive and the top-3 influential journals in this field. USA, Japan and China have contributed to over half of the publication productivity, but European countries seemed to publish more influential articles. It seemed that developed countries/regions tended to produce more articles and more influential articles, and international collaborations mainly occurred among USA, Europe and eastern Asia. Publications concerning surgical management was the major topic, followed by radiographic assessment and epidemiology for this field. Surgical management especially minimally invasive technique for lumbar spondylolisthesis were the recent hotspots over the past 5 years.ConclusionsThe study successfully summarized the productivity and impact of different entities, which should benefit the journal selection and pursuit of international collaboration for researcher who were interested in the field of lumbar spondylolisthesis. Additionally, the current study may encourage more researchers joining in the field and somewhat inform their research direction in the future.

Project description:BackgroundDespite existing research on text mining and machine learning for title and abstract screening, the role of machine learning within systematic literature reviews (SLRs) for health technology assessment (HTA) remains unclear given lack of extensive testing and of guidance from HTA agencies. We sought to address two knowledge gaps: to extend ML algorithms to provide a reason for exclusion-to align with current practices-and to determine optimal parameter settings for feature-set generation and ML algorithms.MethodsWe used abstract and full-text selection data from five large SLRs (n = 3089 to 12,769 abstracts) across a variety of disease areas. Each SLR was split into training and test sets. We developed a multi-step algorithm to categorize each citation into the following categories: included; excluded for each PICOS criterion; or unclassified. We used a bag-of-words approach for feature-set generation and compared machine learning algorithms using support vector machines (SVMs), naïve Bayes (NB), and bagged classification and regression trees (CART) for classification. We also compared alternative training set strategies: using full data versus downsampling (i.e., reducing excludes to balance includes/excludes because machine learning algorithms perform better with balanced data), and using inclusion/exclusion decisions from abstract versus full-text screening. Performance comparisons were in terms of specificity, sensitivity, accuracy, and matching the reason for exclusion.ResultsThe best-fitting model (optimized sensitivity and specificity) was based on the SVM algorithm using training data based on full-text decisions, downsampling, and excluding words occurring fewer than five times. The sensitivity and specificity of this model ranged from 94 to 100%, and 54 to 89%, respectively, across the five SLRs. On average, 75% of excluded citations were excluded with a reason and 83% of these citations matched the reviewers' original reason for exclusion. Sensitivity significantly improved when both downsampling and abstract decisions were used.ConclusionsML algorithms can improve the efficiency of the SLR process and the proposed algorithms could reduce the workload of a second reviewer by identifying exclusions with a relevant PICOS reason, thus aligning with HTA guidance. Downsampling can be used to improve study selection, and improvements using full-text exclusions have implications for a learn-as-you-go approach.

Project description:The molecular basis of diabetes mellitus is yet to be fully elucidated. We aimed to identify the most frequently reported and differential expressed genes (DEGs) in diabetes by using bioinformatics approaches. Text mining was used to screen 40,225 article abstracts from diabetes literature. These studies highlighted 5939 diabetes-related genes spread across 22 human chromosomes, with 112 genes mentioned in more than 50 studies. Among these genes, HNF4A, PPARA, VEGFA, TCF7L2, HLA-DRB1, PPARG, NOS3, KCNJ11, PRKAA2, and HNF1A were mentioned in more than 200 articles. These genes are correlated with the regulation of glycogen and polysaccharide, adipogenesis, AGE/RAGE, and macrophage differentiation. Three datasets (44 patients and 57 controls) were subjected to gene expression analysis. The analysis revealed 135 significant DEGs, of which CEACAM6, ENPP4, HDAC5, HPCAL1, PARVG, STYXL1, VPS28, ZBTB33, ZFP37 and CCDC58 were the top 10 DEGs. These genes were enriched in aerobic respiration, T-cell antigen receptor pathway, tricarboxylic acid metabolic process, vitamin D receptor pathway, toll-like receptor signaling, and endoplasmic reticulum (ER) unfolded protein response. The results of text mining and gene expression analyses used as attribute values for machine learning (ML) analysis. The decision tree, extra-tree regressor and random forest algorithms were used in ML analysis to identify unique markers that could be used as diabetes diagnosis tools. These algorithms produced prediction models with accuracy ranges from 0.6364 to 0.88 and overall confidence interval (CI) of 95%. There were 39 biomarkers that could distinguish diabetic and non-diabetic patients, 12 of which were repeated multiple times. The majority of these genes are associated with stress response, signalling regulation, locomotion, cell motility, growth, and muscle adaptation. Machine learning algorithms highlighted the use of the HLA-DQB1 gene as a biomarker for diabetes early detection. Our data mining and gene expression analysis have provided useful information about potential biomarkers in diabetes.

Project description:BackgroundUnintentional injury is the leading cause of death in young children. Emergency department (ED) diagnoses are a useful source of information for injury epidemiological surveillance purposes. However, ED data collection systems often use free-text fields to report patient diagnoses. Machine learning techniques (MLTs) are powerful tools for automatic text classification. The MLT system is useful to improve injury surveillance by speeding up the manual free-text coding tasks of ED diagnoses.ObjectiveThis research aims to develop a tool for automatic free-text classification of ED diagnoses to automatically identify injury cases. The automatic classification system also serves for epidemiological purposes to identify the burden of pediatric injuries in Padua, a large province in the Veneto region in the Northeast Italy.MethodsThe study includes 283,468 pediatric admissions between 2007 and 2018 to the Padova University Hospital ED, a large referral center in Northern Italy. Each record reports a diagnosis by free text. The records are standard tools for reporting patient diagnoses. An expert pediatrician manually classified a randomly extracted sample of approximately 40,000 diagnoses. This study sample served as the gold standard to train an MLT classifier. After preprocessing, a document-term matrix was created. The machine learning classifiers, including decision tree, random forest, gradient boosting method (GBM), and support vector machine (SVM), were tuned by 4-fold cross-validation. The injury diagnoses were classified into 3 hierarchical classification tasks, as follows: injury versus noninjury (task A), intentional versus unintentional injury (task B), and type of unintentional injury (task C), according to the World Health Organization classification of injuries.ResultsThe SVM classifier achieved the highest performance accuracy (94.14%) in classifying injury versus noninjury cases (task A). The GBM method produced the best results (92% accuracy) for the unintentional and intentional injury classification task (task B). The highest accuracy for the unintentional injury subclassification (task C) was achieved by the SVM classifier. The SVM, random forest, and GBM algorithms performed similarly against the gold standard across different tasks.ConclusionsThis study shows that MLTs are promising techniques for improving epidemiological surveillance, allowing for the automatic classification of pediatric ED free-text diagnoses. The MLTs revealed a suitable classification performance, especially for general injuries and intentional injury classification. This automatic classification could facilitate the epidemiological surveillance of pediatric injuries by also reducing the health professionals' efforts in manually classifying diagnoses for research purposes.

Project description:Introduction. Suicide has become a serious worldwide epidemic. Early detection of individual suicide risk in population is important for reducing suicide rates. Traditional methods are ineffective in identifying suicide risk in time, suggesting a need for novel techniques. This paper proposes to detect suicide risk on social media using a Chinese suicide dictionary. Methods. To build the Chinese suicide dictionary, eight researchers were recruited to select initial words from 4,653 posts published on Sina Weibo (the largest social media service provider in China) and two Chinese sentiment dictionaries (HowNet and NTUSD). Then, another three researchers were recruited to filter out irrelevant words. Finally, remaining words were further expanded using a corpus-based method. After building the Chinese suicide dictionary, we tested its performance in identifying suicide risk on Weibo. First, we made a comparison of the performance in both detecting suicidal expression in Weibo posts and evaluating individual levels of suicide risk between the dictionary-based identifications and the expert ratings. Second, to differentiate between individuals with high and non-high scores on self-rating measure of suicide risk (Suicidal Possibility Scale, SPS), we built Support Vector Machines (SVM) models on the Chinese suicide dictionary and the Simplified Chinese Linguistic Inquiry and Word Count (SCLIWC) program, respectively. After that, we made a comparison of the classification performance between two types of SVM models. Results and Discussion. Dictionary-based identifications were significantly correlated with expert ratings in terms of both detecting suicidal expression (r = 0.507) and evaluating individual suicide risk (r = 0.455). For the differentiation between individuals with high and non-high scores on SPS, the Chinese suicide dictionary (t1: F 1 = 0.48; t2: F 1 = 0.56) produced a more accurate identification than SCLIWC (t1: F 1 = 0.41; t2: F 1 = 0.48) on different observation windows. Conclusions. This paper confirms that, using social media, it is possible to implement real-time monitoring individual suicide risk in population. Results of this study may be useful to improve Chinese suicide prevention programs and may be insightful for other countries.

Dataset Information

Assessing Suicide Risk and Emotional Distress in Chinese Social Media: A Text Mining and Machine Learning Study.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets