AI Naturalists Might Hold the Key to Unlocking Biodiversity Data in Social Media Imagery.
Ontology highlight
ABSTRACT:
SUBMITTER: August TA
PROVIDER: S-EPMC7660428 | biostudies-literature | 2020 Oct
REPOSITORIES: biostudies-literature
Publications
AI Naturalists Might Hold the Key to Unlocking Biodiversity Data in Social Media Imagery.
August Tom A TA Pescott Oliver L OL Joly Alexis A Bonnet Pierre P
Patterns (New York, N.Y.) 20201009 7
The increasing availability of digital images, coupled with sophisticated artificial intelligence (AI) techniques for image classification, presents an exciting opportunity for biodiversity researchers to create new datasets of species observations. We investigated whether an AI plant species classifier could extract previously unexploited biodiversity data from social media photos (Flickr). We found over 60,000 geolocated images tagged with the keyword "flower" across an urban and rural locatio ...[more]
Similar Datasets
Project description:Targeting of location-specific aid for the U.S. opioid epidemic is difficult due to our inability to accurately predict changes in opioid mortality across heterogeneous communities. AI-based language analyses, having recently shown promise in cross-sectional (between-community) well-being assessments, may offer a way to more accurately longitudinally predict community-level overdose mortality. Here, we develop and evaluate, TROP (Transformer for Opiod Prediction), a model for community-specific trend projection that uses community-specific social media language along with past opioid-related mortality data to predict future changes in opioid-related deaths. TOP builds on recent advances in sequence modeling, namely transformer networks, to use changes in yearly language on Twitter and past mortality to project the following year's mortality rates by county. Trained over five years and evaluated over the next two years TROP demonstrated state-of-the-art accuracy in predicting future county-specific opioid trends. A model built using linear auto-regression and traditional socioeconomic data gave 7% error (MAPE) or within 2.93 deaths per 100,000 people on average; our proposed architecture was able to forecast yearly death rates with less than half that error: 3% MAPE and within 1.15 per 100,000 people.
| S-EPMC9992514 | biostudies-literature
Project description: Not available
| S-EPMC11344235 | biostudies-literature
Project description:IntroductionThe integration of AI in architectural design represents a significant shift toward creating emotionally resonant spaces. This research investigates AI's ability to evoke specific emotional responses through architectural imagery and examines the impact of professional training on emotional interpretation.MethodsWe utilized Midjourney AI software to generate images based on direct and metaphorical prompts across two architectural settings: home interiors and museum exteriors. A survey was designed to capture participants' emotional responses to these images, employing a scale that rated their immediate emotional reaction. The study involved 789 university students, categorized into architecture majors (Group A) and non-architecture majors (Group B), to explore differences in emotional perception attributable to educational background.ResultsFindings revealed that AI is particularly effective in depicting joy, especially in interior settings. However, it struggles to accurately convey negative emotions, indicating a gap in AI's emotional range. Architecture students exhibited a greater sensitivity to emotional nuances in the images compared to non-architecture students, suggesting that architectural training enhances emotional discernment. Notably, the study observed minimal differences in the perception of emotions between direct and metaphorical prompts among architecture students, indicating a consistent emotional interpretation across prompt types.ConclusionAI holds significant promise in creating spaces that resonate on an emotional level, particularly in conveying positive emotions like joy. The study contributes to the understanding of AI's role in architectural design, emphasizing the importance of emotional intelligence in creating spaces that reflect human experiences. Future research should focus on expanding AI's emotional range and further exploring the impact of architectural training on emotional perception.
| S-EPMC10963507 | biostudies-literature
Project description:Shaped by human movement, place connectivity is quantified by the strength of spatial interactions among locations. For decades, spatial scientists have researched place connectivity, applications, and metrics. The growing popularity of social media provides a new data stream where spatial social interaction measures are largely devoid of privacy issues, easily assessable, and harmonized. In this study, we introduced a place connectivity index (PCI) based on spatial interactions among places revealed by geotagged tweets as a multiscale, spatiotemporal-continuous, and easy-to-implement measurement. The proposed PCI, established and demonstrated at the US county level, exhibits a strong positive association with SafeGraph population movement records (10 percent penetration in the US population) and Facebook social connectedness index (SCI), a popular connectivity index based on social networks. We found that PCI has a strong state boundary effect and that it generally follows the distance decay effect, although this force is weaker in more urbanized counties with a denser population. Our investigation further suggests that PCI has great potential in addressing real-world problems that require place connectivity knowledge, exemplified with two applications: 1) modeling the spatial spread of a contagious disease (e.g., COVID-19), and 2) modeling hurricane evacuation destination choices. The methodological and contextual knowledge of PCI, together with the launched visualization platform and data sharing capability, is expected to support research fields requiring knowledge in human spatial interactions.
| S-EPMC7872361 | biostudies-literature
Project description:Suicide is not only an individual phenomenon, but it is also influenced by social and environmental factors. With the high suicide rate and the abundance of social media data in South Korea, we have studied the potential of this new medium for predicting completed suicide at the population level. We tested two social media variables (suicide-related and dysphoria-related weblog entries) along with classical social, economic and meteorological variables as predictors of suicide over 3 years (2008 through 2010). Both social media variables were powerfully associated with suicide frequency. The suicide variable displayed high variability and was reactive to celebrity suicide events, while the dysphoria variable showed longer secular trends, with lower variability. We interpret these as reflections of social affect and social mood, respectively. In the final multivariate model, the two social media variables, especially the dysphoria variable, displaced two classical economic predictors - consumer price index and unemployment rate. The prediction model developed with the 2-year training data set (2008 through 2009) was validated in the data for 2010 and was robust in a sensitivity analysis controlling for celebrity suicide effects. These results indicate that social media data may be of value in national suicide forecasting and prevention.
| S-EPMC3632511 | biostudies-literature
Project description:The data is for empirical studies derived from a related research article about space tourism [1], which is a conceptual article with a different aim of economic measurement scale. Most space tourism research is conceptual because data from the nascent industry is limited [2]. Thus this data is constrictive for conducting empirical studies to contribute to quantitative analysis in space tourism [3]. Data from this study were collected by recruiting 361 respondents through snowball and convenient sampling targeted to samples interested in space tourism; 339 responses were adopted after the valid screening of missing data or data bias [4]. Targeted groups of potential space tourism customers were investigated to collect data through a designed questionnaire on the platform, Wenjuanxing, with a majority population database providing functions equivalent to Amazon Mechanical Turk [2]. The reliability and validity of all constructs showed that the questionnaire was proper for measurement [3]. Data analysis applied the structural equation model with Mplus to examine the CFA model and research hypothesis. Structural equation modeling was used to conduct the hypotheses test and model fitness through the statistical tool Mplus. Results imply that the data is suitable for conducting replication studies. To enlighten space tourism emergence studies, this data shows its importance for further research models [5].
| S-EPMC10051023 | biostudies-literature
Project description:Depression has robust natural language correlates and can increasingly be measured in language using predictive models. However, despite evidence that language use varies as a function of individual demographic features (e.g., age, gender), previous work has not systematically examined whether and how depression's association with language varies by race. We examine how race moderates the relationship between language features (i.e., first-person pronouns and negative emotions) from social media posts and self-reported depression, in a matched sample of Black and White English speakers in the United States. Our findings reveal moderating effects of race: While depression severity predicts I-usage in White individuals, it does not in Black individuals. White individuals use more belongingness and self-deprecation-related negative emotions. Machine learning models trained on similar amounts of data to predict depression severity performed poorly when tested on Black individuals, even when they were trained exclusively using the language of Black individuals. In contrast, analogous models tested on White individuals performed relatively well. Our study reveals surprising race-based differences in the expression of depression in natural language and highlights the need to understand these effects better, especially before language-based models for detecting psychological phenomena are integrated into clinical practice.
| S-EPMC10998627 | biostudies-literature
Project description:BackgroundAI models have shown promise in performing many medical imaging tasks. However, our ability to explain what signals these models have learned is severely lacking. Explanations are needed in order to increase the trust of doctors in AI-based models, especially in domains where AI prediction capabilities surpass those of humans. Moreover, such explanations could enable novel scientific discovery by uncovering signals in the data that aren't yet known to experts.MethodsIn this paper, we present a workflow for generating hypotheses to understand which visual signals in images are correlated with a classification model's predictions for a given task. This approach leverages an automatic visual explanation algorithm followed by interdisciplinary expert review. We propose the following 4 steps: (i) Train a classifier to perform a given task to assess whether the imagery indeed contains signals relevant to the task; (ii) Train a StyleGAN-based image generator with an architecture that enables guidance by the classifier ("StylEx"); (iii) Automatically detect, extract, and visualize the top visual attributes that the classifier is sensitive towards. For visualization, we independently modify each of these attributes to generate counterfactual visualizations for a set of images (i.e., what the image would look like with the attribute increased or decreased); (iv) Formulate hypotheses for the underlying mechanisms, to stimulate future research. Specifically, present the discovered attributes and corresponding counterfactual visualizations to an interdisciplinary panel of experts so that hypotheses can account for social and structural determinants of health (e.g., whether the attributes correspond to known patho-physiological or socio-cultural phenomena, or could be novel discoveries).FindingsTo demonstrate the broad applicability of our approach, we present results on eight prediction tasks across three medical imaging modalities-retinal fundus photographs, external eye photographs, and chest radiographs. We showcase examples where many of the automatically-learned attributes clearly capture clinically known features (e.g., types of cataract, enlarged heart), and demonstrate automatically-learned confounders that arise from factors beyond physiological mechanisms (e.g., chest X-ray underexposure is correlated with the classifier predicting abnormality, and eye makeup is correlated with the classifier predicting low hemoglobin levels). We further show that our method reveals a number of physiologically plausible, previously-unknown attributes based on the literature (e.g., differences in the fundus associated with self-reported sex, which were previously unknown).InterpretationOur approach enables hypotheses generation via attribute visualizations and has the potential to enable researchers to better understand, improve their assessment, and extract new knowledge from AI-based models, as well as debug and design better datasets. Though not designed to infer causality, importantly, we highlight that attributes generated by our framework can capture phenomena beyond physiology or pathophysiology, reflecting the real world nature of healthcare delivery and socio-cultural factors, and hence interdisciplinary perspectives are critical in these investigations. Finally, we will release code to help researchers train their own StylEx models and analyze their predictive tasks of interest, and use the methodology presented in this paper for responsible interpretation of the revealed attributes.FundingGoogle.
| S-EPMC10993140 | biostudies-literature
Project description:Using social media for health promotion is an innovative and emerging approach but remains relatively unexplored in cancer screening. Uptake of colorectal cancer (CRC) screening remains low and standard methods of reaching out are expensive with limited impact. The objective of this study is to conduct a cluster randomized controlled trial (RCT) to test the effectiveness of social media messages for CRC screening on screening intention (primary outcome). The results of this trial will be of interest to Cancer Care Ontario and are likely to be taken up by other screening programs looking for innovative and novel ways to increase screening participation. The study results will be easily translatable identifying the most compelling CRC screening messages while the approach can easily be translated to other cancer disease sites with screening programs.
| 2332186 | ecrin-mdr-crc
Project description:As e-cigarette use rapidly increases in popularity, data from online social systems (Twitter, Instagram, Google Web Search) can be used to capture and describe the social and environmental context in which individuals use, perceive, and are marketed this tobacco product. Social media data may serve as a massive focus group where people organically discuss e-cigarettes unprimed by a researcher, without instrument bias, captured in near real time and at low costs.This study documents e-cigarette-related discussions on Twitter, describing themes of conversations and locations where Twitter users often discuss e-cigarettes, to identify priority areas for e-cigarette education campaigns. Additionally, this study demonstrates the importance of distinguishing between social bots and human users when attempting to understand public health-related behaviors and attitudes.E-cigarette-related posts on Twitter (N=6,185,153) were collected from December 24, 2016, to April 21, 2017. Techniques drawn from network science were used to determine discussions of e-cigarettes by describing which hashtags co-occur (concept clusters) in a Twitter network. Posts and metadata were used to describe where geographically e-cigarette-related discussions in the United States occurred. Machine learning models were used to distinguish between Twitter posts reflecting attitudes and behaviors of genuine human users from those of social bots. Odds ratios were computed from 2x2 contingency tables to detect if hashtags varied by source (social bot vs human user) using the Fisher exact test to determine statistical significance.Clusters found in the corpus of hashtags from human users included behaviors (eg, #vaping), vaping identity (eg, #vapelife), and vaping community (eg, #vapenation). Additional clusters included products (eg, #eliquids), dual tobacco use (eg, #hookah), and polysubstance use (eg, #marijuana). Clusters found in the corpus of hashtags from social bots included health (eg, #health), smoking cessation (eg, #quitsmoking), and new products (eg, #ismog). Social bots were significantly more likely to post hashtags that referenced smoking cessation and new products compared to human users. The volume of tweets was highest in the Mid-Atlantic (eg, Pennsylvania, New Jersey, Maryland, and New York), followed by the West Coast and Southwest (eg, California, Arizona and Nevada).Social media data may be used to complement and extend the surveillance of health behaviors including tobacco product use. Public health researchers could harness these data and methods to identify new products or devices. Furthermore, findings from this study demonstrate the importance of distinguishing between Twitter posts from social bots and humans when attempting to understand attitudes and behaviors. Social bots may be used to perpetuate the idea that e-cigarettes are helpful in cessation and to promote new products as they enter the marketplace.
| S-EPMC5752967 | biostudies-literature