A5023662337- Topic Modeling
- Biomedical Text Mining and Ontologies
- Semantic Web and Ontologies
- Sentiment Analysis and Opinion Mining
- Advanced Text Analysis Techniques
- Bioinformatics and Genomic Networks
- Natural Language Processing Techniques
- Nutritional Studies and Diet
- Social Media in Health Education
- Mental Health via Writing
- Advanced Computational Techniques in Science and Engineering
- Educational Technology and Optimization
- Media Influence and Health
- Reading and Literacy Development
- Machine Learning and Algorithms
- Child and Animal Learning Development
- Sensory Analysis and Statistical Methods
- Microbial Natural Products and Biosynthesis
- Advanced Signal Processing Techniques
- Access Control and Trust
- Service-Learning and Community Engagement
- Olfactory and Sensory Function Studies
- Smoking Behavior and Cessation
- Behavioral Health and Interventions
- Neurobiology of Language and Bilingualism
University of Arizona
2014-2021
Concordia University
2021
Concordia University
2021
University of Illinois Urbana-Champaign
2021
IBM Research - Thomas J. Watson Research Center
2021
University of Tennessee at Knoxville
2009
We investigate the predictive power behind language of food on social media. collect a corpus over three million food-related posts from Twitter and demonstrate that many latent population characteristics can be directly predicted this data: overweight rate, diabetes political leaning, home geographical location authors. For all tasks, our language-based models significantly outperform majority-class baselines. Performance is further improved with more complex natural processing, such as...
PubMed, a repository and search engine for biomedical literature, now indexes >1 million articles each year. This exceeds the processing capacity of human domain experts, limiting our ability to truly understand many diseases. We present Reach, system automated, large-scale machine reading papers that can extract mechanistic descriptions biological processes with relatively high precision at throughput. demonstrate combining extracted pathway fragments existing data analysis algorithms rely...
The risk perception attitude (RPA) framework was tested as a message tailoring strategy to encourage diabetes screening. Participants (N = 602) were first categorized into one of four RPA groups based on their and efficacy perceptions then randomly assigned receive that matched RPA, mismatched or control message. receiving reported greater intentions engage in self-protective behavior than participants who received the results also showed differences attitudes behavioral across groups....
Causal precedence between biochemical interactions is crucial in the biomedical domain, because it transforms collections of individual interactions, e.g., bindings and phosphorylations, into causal mechanisms needed to inform meaningful search inference.Here, we analyze domain as distinct from open-domain, temporal precedence.First, describe a novel, hand-annotated text corpus domain.Second, use this investigate battery models precedence, covering rule-based, feature-based, latent...
Software designed to accurately estimate food calories from still images could help users and health professionals identify dietary patterns choices associated with risks more effectively. However, calorie estimation is difficult, no publicly available software can do so while minimizing the burden data collection analysis.The aim of this study was determine accuracy crowdsourced annotations content in quantify sources bias noise as a function respondent characteristics qualities (eg, energy...
George C. G. Barbosa, Zechy Wong, Gus Hahn-Powell, Dane Bell, Rebecca Sharp, Marco A. Valenzuela-Escárcega, Mihai Surdeanu. Proceedings of the 2019 Conference North American Chapter Association for Computational Linguistics (Demonstrations). 2019.
This work explores the detection of individuals' risk type 2 diabetes mellitus (T2DM) directly from their social media (Twitter) activity. Our approach extends a deep learning architecture with several contributions: following previous observations that language use differs by gender, it captures and uses gender information through domain adaptation; recency posts under hypothesis more recent are representative an individual's current status; and, lastly, demonstrates in this scenario where...
We propose an approach for biomedical information extraction that marries the advantages of machine learning models, e.g., directly from data, with benefits rule-based approaches, interpretability. Our starts by training a feature-based statistical model, then converts this model to variant converting its features rules, and “snapping grid” feature weights discrete votes. In doing so, our proposal takes advantage large body work in learning, but it produces interpretable which can be edited...
We describe a strategy for the acquisition of training data necessary to build social-media-driven early detection system individuals at risk (preventable) type 2 diabetes mellitus (T2DM). The uses game-like quiz with and questions acquired semi-automatically from Twitter. are designed inspire participant engagement collect relevant train public-health model applied individuals. Prior systems use social media such as Twitter predict obesity (a factor T2DM) operate on entire communities...
We describe challenges and advantages unique to coreference resolution in the biomedical domain, a sieve-based architecture that leverages domain knowledge for both entity event resolution. Domain-general algorithms perform poorly on documents, because cues they rely such as gender are largely absent this do not encode domain-specific number type of participants required chemical reactions. Moreover, it is difficult directly into most rule-based. Our rule-based uses sequentially applied...
Causal precedence between biochemical interactions is crucial in the biomedical domain, because it transforms collections of individual interactions, e.g., bindings and phosphorylations, into causal mechanisms needed to inform meaningful search inference. Here, we analyze domain as distinct from open-domain, temporal precedence. First, describe a novel, hand-annotated text corpus domain. Second, use this investigate battery models precedence, covering rule-based, feature-based, latent...
We propose an approach for biomedical information extraction that marries the advantages of machine learning models, e.g., directly from data, with benefits rule-based approaches, interpretability. Our starts by training a feature-based statistical model, then converts this model to variant converting its features rules, and "snapping grid" feature weights discrete votes. In doing so, our proposal takes advantage large body work in learning, but it produces interpretable which can be edited...