Background: Mobile augmented reality (MAR) apps offer potential support for emergency responders in rural areas. Objective: In this report, we described lessons learned from the development process of augmented reality (AR) Farm Mapping to Assist, Protect and Prepare Emergency Responders (MAPPER), a MAR app that provides emergency responders onsite information about the agricultural operation they enter. Methods: Cross-platform frameworks were used to create AR MAPPER to accommodate budget constraints and overcome issues with markerless MAR technologies. Although the single codebase and Web technologies streamlined development, cross-device hardware limitations impacted location accuracy, lengthened the development cycle, and required regular updates to third-party libraries. Results: A hybrid development approach of using Web-based technologies with native tie-ins for specialized components and enhanced performance cut time and costs. This also led to consistency across multiple platforms and ensured that there is only a single set of source files to modify for Android and iPhone operating systems. Meanwhile, active development was delayed by some major hurdles. Apple and Google both released new versions of their operating systems, and the Wikitude framework issued four major updates, each of which brought with it some important enhancements and also led to some new issues. Conclusions: Developers should consider single platform native development to benefit from platform-specific MAR implementations and to avoid development, testing, and maintenance costs associated with cross-platform implementation. Emergency response organizations may be more likely to utilize a single platform across the devices used by their command staff. This also reduces the benefits of cross-platform development. Furthermore, providing map-based, non-AR cross-platform apps for landowners, farmers, and ranchers would help improve and maintain data quality, which is crucial for the utility and user experience of MAR apps.

AgInjuryNews.org is a news report-based, online sentinel surveillance dataset that has provided publicly available news and media reports of agricultural injuries since early 2015. In the 6 years since its inception, AgInjuryNews.org has hosted 12,897 unique visitors and has collected 997 user account registrations. New users from geographic areas home to NIOSH-funded agricultural research centres were most prominent, with these centres returning in larger numbers, comparatively. Users were acquired mostly through web searches, collaborations with other agencies, and paid Facebook.com advertisements. Paid advertisements recruited 3792 visitors; however, retention, registrations, and on-site engagement from this source was low. This analysis shows that data consumption on AgInjuryNews.org is steadily growing. Similar self-hosted programs that provide data or digital resources to agricultural safety and health stakeholders should consider the integration of auditing and analytics tracking, including user registrations.

The traditional approach to curating and disseminating information about agricultural injuries relies heavily on manual input and review, resulting in a labor-intensive process. While the unstructured nature of the material traditionally requires human reviewers, the recent proliferation of Large Language Models (LLMs) has introduced the potential for automation. This study investigates the feasibility and implications of filling the role of a human reviewer with an LLM in analyzing information about agricultural injuries from news articles and investigation reports. Multiple language models were tested for accuracy in extracting relevant incident and victim information, and these models include OpenAI’s ChatGPT 3.5 and 4 and an open-source fine-tuned version of Llama 2. To measure accuracy, each LLM was given prompts to gather relevant data from a set of randomly selected online news articles already cataloged by human reviewers, such as the use of drugs or alcohol, time of day, or other information about the victim(s). Results showed that the fine-tuned Llama2 was the most proficient model with an average accuracy of 93% and some categories reaching 100%. ChatGPT-4 also performed well with around 90% accuracy. Additionally, we found that the fine-tuned Llama2 model was somewhat proficient in coding injuries using the OIICS classification scheme, achieving 48% accuracy when predicting the first digit. Though none of the models are perfectly accurate, the methodology and results prove that LLMs are promising in streamlining workflows in order to reduce human and financial resources and increase the efficiency of data analysis.

Purpose: Enhanced monitoring of user acquisition and online use patterns throughout the AgInjuryNews.org data repository is imperative toward understanding overall website usage, trends, and behavior. This study examines how implementing refined, modernized analytical toolsets can provide more granular and versatile tracking of data and, in turn, reveal new focus areas to maximize effectiveness of data dissemination to end users. Methods: Collection of analytical data was expanded from its original rudimentary form to include advanced capabilities of Google Tag Manager (GTM), beginning October 18, 2019. This solution allows real-time, adaptable tracking of all operations and actions across the website. GTM also provides enhanced tie-ins with social media referrals and advertising campaign events. By using GTM as the new baseline analytics foundation, it becomes possible for ad-hoc integration of over 50 third-party analytical components and services. Results/Findings: The post-GTM dataset has not yet matured; more time and website traffic is needed to develop a comprehensive picture of overall usage behavior to optimize the content and user experience. Preliminary analysis has led to the discovery of common user actions including specialized filtering of injury data by age group and number of victims, as well as common search queries including “Tractor” and “Silo.” We have also observed retention deficits by users who arrive initially through social media campaigns versus organic return-visitor traffic. Practical application: Evaluating the GTM-based analytical dataset of AgInjuryNews.org allows for a deeper understanding of trends and behaviors relating to agricultural injury research. The findings help determine strengths and weaknesses in our existing web interface so refinements can be made to improve user retention and website functionality. For instance, by studying commonly searched topics and filter options we can better target social media ads and encourage more organic return-visitor traffic. This knowledge allows us to refine future campaigns and tailor first-time visitor experiences to better engage and retain our prospective audience.

UNSTRUCTURED Agriculture, Fishing, and Forestry is the most dangerous occupation in the United States. AgInjuryNews.org is a news report-based online sentinel surveillance project that has collected news and media reports of agricultural injuries since early 2015. We aimed to provide a brief report on visitor acquisition, retention, and usage as well as monitor the efficacy of a Facebook.com social media marketing campaign. Since its inception, AgInjuryNews.org hosted 9,281 unique visitors. New users from major cities and areas housing agricultural research centers were most prominent with agricultural centers returning in larger numbers, comparatively. Users were acquired mostly through web searches, partnerships with other agencies, and paid Facebook.com advertisements. These paid advertisements recruited nearly 4000 new users, however retention and engagement was low. This analysis shows that AgInjuryNews.org is steadily growing and has further informed how to continue to recruit and engage new users.

BACKGROUND Mobile augmented reality (MAR) apps offer potential support for emergency responders in rural areas. OBJECTIVE In this report, we described lessons learned from the development process of augmented reality (AR) Farm Mapping to Assist, Protect and Prepare Emergency Responders (MAPPER), a MAR app that provides emergency responders onsite information about the agricultural operation they enter. METHODS Cross-platform frameworks were used to create AR MAPPER to accommodate budget constraints and overcome issues with markerless MAR technologies. Although the single codebase and Web technologies streamlined development, cross-device hardware limitations impacted location accuracy, lengthened the development cycle, and required regular updates to third-party libraries. RESULTS A hybrid development approach of using Web-based technologies with native tie-ins for specialized components and enhanced performance cut time and costs. This also led to consistency across multiple platforms and ensured that there is only a single set of source files to modify for Android and iPhone operating systems. Meanwhile, active development was delayed by some major hurdles. Apple and Google both released new versions of their operating systems, and the Wikitude framework issued four major updates, each of which brought with it some important enhancements and also led to some new issues. CONCLUSIONS Developers should consider single platform native development to benefit from platform-specific MAR implementations and to avoid development, testing, and maintenance costs associated with cross-platform implementation. Emergency response organizations may be more likely to utilize a single platform across the devices used by their command staff. This also reduces the benefits of cross-platform development. Furthermore, providing map-based, non-AR cross-platform apps for landowners, farmers, and ranchers would help improve and maintain data quality, which is crucial for the utility and user experience of MAR apps.