Background: Prescription medication (PM) misuse/abuse has emerged as a national crisis in the United States, and social media has been suggested as a potential resource for performing active monitoring. However, automating a social media-based monitoring system is challenging—requiring advanced natural language processing (NLP) and machine learning methods. In this paper, we describe the development and evaluation of automatic text classification models for detecting self-reports of PM abuse from Twitter.Methods: We experimented with state-of-the-art bi-directional transformer-based language models, which utilize tweet-level representations that enable transfer learning (e.g., BERT, RoBERTa, XLNet, AlBERT, and DistilBERT), proposed fusion-based approaches, and compared the developed models with several traditional machine learning, including deep learning, approaches. Using a public dataset, we evaluated the performances of the classifiers on their abilities to classify the non-majority “abuse/misuse” class.Results: Our proposed fusion-based model performs significantly better than the best traditional model (F₁-score [95% CI]: 0.67 [0.64–0.69] vs. 0.45 [0.42–0.48]). We illustrate, via experimentation using varying training set sizes, that the transformer-based models are more stable and require less annotated data compared to the other models. The significant improvements achieved by our best-performing classification model over past approaches makes it suitable for automated continuous monitoring of nonmedical PM use from Twitter.Conclusions: BERT, BERT-like and fusion-based models outperform traditional machine learning and deep learning models, achieving substantial improvements over many years of past research on the topic of prescription medication misuse/abuse classification from social media, which had been shown to be a complex task due to the unique ways in which information about nonmedical use is presented. Several challenges associated with the lack of context and the nature of social media language need to be overcome to further improve BERT and BERT-like models. These experimental driven challenges are represented as potential future research directions.

The COVID-19 pandemic is the most devastating public health crisis in at least a century and has affected the lives of billions of people worldwide in unprecedented ways. Compared to pandemics of this scale in the past, societies are now equipped with advanced technologies that can mitigate the impacts of pandemics if utilized appropriately. However, opportunities are currently not fully utilized, particularly at the intersection of data science and health. Health-related big data and technological advances have the potential to significantly aid the fight against such pandemics, including the current pandemic’s ongoing and long-term impacts. Specifically, the field of natural language processing (NLP) has enormous potential at a time when vast amounts of text-based data are continuously generated from a multitude of sources, such as health/hospital systems, published medical literature, and social media. Effectively mitigating the impacts of the pandemic requires tackling challenges associated with the application and deployment of NLP systems. In this paper, we review the applications of NLP to address diverse aspects of the COVID-19 pandemic. We outline key NLP-related advances on a chosen set of topics reported in the literature and discuss the opportunities and challenges associated with applying NLP during the current pandemic and future ones. These opportunities and challenges can guide future research aimed at improving the current health and social response systems and pandemic preparedness.

Intimate partner violence (IPV) is a preventable public health problem that affects millions of people worldwide. Approximately one in four women are estimated to be or have been victims of severe violence at some point in their lives, irrespective of age, ethnicity, and economic status. Victims often report IPV experiences on social media, and automatic detection of such reports via machine learning may enable improved surveillance and targeted distribution of support and/or interventions for those in need. However, no artificial intelligence systems for automatic detection currently exists, and we attempted to address this research gap. We collected posts from Twitter using a list of IPV-related keywords, manually reviewed subsets of retrieved posts, and prepared annotation guidelines to categorize tweets into IPV-report or non-IPV-report. We annotated 6,348 tweets in total, with the inter-annotator agreement (IAA) of 0.86 (Cohen's kappa) among 1,834 double-annotated tweets. The class distribution in the annotated dataset was highly imbalanced, with only 668 posts (~11%) labeled as IPV-report. We then developed an effective natural language processing model to identify IPV-reporting tweets automatically. The developed model achieved classification F₁-scores of 0.76 for the IPV-report class and 0.97 for the non-IPV-report class. We conducted post-classification analyses to determine the causes of system errors and to ensure that the system did not exhibit biases in its decision making, particularly with respect to race and gender. Our automatic model can be an essential component for a proactive social media-based intervention and support framework, while also aiding population-level surveillance and large-scale cohort studies.

Pretrained contextual language models proposed in the recent past have been reported to achieve state-of-the-art performances in many natural language processing (NLP) tasks, including those involving health-related social media data. We sought to evaluate the effectiveness of different pretrained transformer-based models for social media-based health-related text classification tasks. An additional objective was to explore and propose effective pretraining strategies to improve machine learning performance on such datasets and tasks. We benchmarked six transformer-based models that were pretrained with texts from different domains and sources—BERT, RoBERTa, BERTweet, TwitterBERT, BioClinical_BERT, and BioBERT—on 22 social media-based health-related text classification tasks. For the top-performing models, we explored the possibility of further boosting performance by comparing several pretraining strategies: domain-adaptive pretraining (DAPT), source-adaptive pretraining (SAPT), and a novel approach called topic specific pretraining (TSPT). We also attempted to interpret the impacts of distinct pretraining strategies by visualizing document-level embeddings at different stages of the training process. RoBERTa outperformed BERTweet on most tasks, and better than others. BERT, TwitterBERT, BioClinical_BERT and BioBERT consistently underperformed. For pretraining strategies, SAPT performed better or comparable to the off-the-shelf models, and significantly outperformed DAPT. SAPT + TSPT showed consistently high performance, with statistically significant improvement in three tasks. Our findings demonstrate that RoBERTa and BERTweet are excellent off-the-shelf models for health-related social media text classification, and extended pretraining using SAPT and TSPT can further improve performance.

Americans bear a high chronic stress burden, particularly during the COVID-19 pandemic. Although social media have many strengths to complement the weaknesses of conventional stress measures, including surveys, they have been rarely utilized to detect individuals self-reporting chronic stress. Thus, this study aimed to develop and evaluate an automatic system on Twitter to identify users who have self-reported chronic stress experiences. Using the Twitter public streaming application programming interface, we collected tweets containing certain stress-related keywords (eg, “chronic,” “constant,” “stress”) and then filtered the data using pre-defined text patterns. We manually annotated tweets with (without) self-report of chronic stress as positive (negative). We trained multiple classifiers and tested them via accuracy and F ₁ score. We annotated 4195 tweets (1560 positives, 2635 negatives), achieving an inter-annotator agreement of 0.83 (Cohen's kappa). The classifier based on Bidirectional Encoder Representation from Transformers performed the best (accuracy of 83.6% [81.0-86.1]), outperforming the second best-performing classifier (support vector machines: 76.4% [73.5-79.3]). The past tweets from the authors of positive tweets contained useful information, including sources and health impacts of chronic stress. Our study demonstrates that users' self-reported chronic stress experiences can be automatically identified on Twitter, which has a high potential for surveillance and large-scale intervention.

The capabilities of natural language processing (NLP) methods have expanded significantly in recent years, and progress has been particularly driven by advances in data science and machine learning. However, NLP is still largely underused in patient-oriented clinical research and care (POCRC). A key reason behind this is that clinical NLP methods are typically developed, optimized, and evaluated with narrowly focused data sets and tasks (eg, those for the detection of specific symptoms in free texts). Such research and development (R&D) approaches may be described as problem oriented, and the developed systems perform specialized tasks well. As standalone systems, however, they generally do not comprehensively meet the needs of POCRC. Thus, there is often a gap between the capabilities of clinical NLP methods and the needs of patient-facing medical experts. We believe that to increase the practical use of biomedical NLP, future R&D efforts need to be broadened to a new research paradigm—one that explicitly incorporates characteristics that are crucial for POCRC. We present our viewpoint about 4 such interrelated characteristics that can increase NLP systems’ suitability for POCRC (3 that represent NLP system properties and 1 associated with the R&D process)—(1) interpretability (the ability to explain system decisions), (2) patient centeredness (the capability to characterize diverse patients), (3) customizability (the flexibility for adapting to distinct settings, problems, and cohorts), and (4) multitask evaluation (the validation of system performance based on multiple tasks involving heterogeneous data sets). By using the NLP task of clinical concept detection as an example, we detail these characteristics and discuss how they may result in the increased uptake of NLP systems for POCRC.