SBIR-STTR Award

USDA seeks to develop tools appropriate for small farms that enhance farm safety and risk management. Farming remains an extremely dangerous occupation: agricultural workers are 33 times more likely to die than workers in other industries (U.S. Bureau of Labor Statistics 2020).Farm-related injuries cost the U.S economy over four billion dollars per year (Leigh et al 2001). VISIMO will develop a machine learning software application to enhance agricultural safety through risk management. VISIMO's model is novel and proprietary and is proven to increase hazard identification by 74% increase risk mitigation by 88% and decrease injury rates by 44% in the tool's original application area of construction job safety analysis. The proposed work is divided into three primary steps: (1) data is gathered risk scores are assigned and this new information is integrated into the model; (2) data is collected in the field using the developed tool and gathered ground-truth data is used to test risk predictions; (3) machine learning is incorporated into the model to significantly increase the accuracy of risk prediction. The proposed risk management tool encompasses the complexity of real-world risk is usable in rural settings without internet access and takes less than 44 seconds to record an observation calculate a risk score and suggest risk-reducing changes. Phase I will result in a proof-of-concept risk model and native web application that is validated on approximately 100-250 variables and is ready for expansion and end-user testing in Phase II. Commercial end users are small and mid-sized farms.

VISIMO proposes to develop a risk management platform, referred to as Demeter, that leverages a custom Machine Learning (ML) model to enhance the efficiency, profitability, and safety of small and mid-sized farms. Demeter will advance the state of farm safety by identifying risks, providing mitigating actions, and offering economic cost-benefit analyses for suggested mitigations. The proposed application connects to agricultural-related manufacturing technology by improving the safety of agricultural practices, increasing productivity, improving operational management, and training farmers and producers. Farming is among the most dangerous occupations in the United States with an annual death rate of 20/100,000 persons (U.S. Bureau of Labor Statistics, 2021). Nationally, workers in Agriculture, Forestry, and Fishing (AFF) industries are up to 33 times more likely to die on the job than workers in other industries (U.S. Bureau of Labor Statistics, 2021). The average annual cost of occupational injuries in agriculture is $8.3B, including both medical costs and lost productivity, and an estimated 167 agricultural workers per day lose productivity due to occupational injuries (Agricultural Safety & Health Council of America, 2015). Farming injuries represent 30% higher cost rates for personal injuries per year than the national average across all industries (Leigh et al., 2001).People under the age of 16 years old are especially impacted by agricultural hazards. Farms, ranches, and other agricultural operations are often both worksites and homes (Reames, L., Jan 31, 2023 VISIMO interview). Of the 731,000 youths working in agriculture, 65% of them work on a family farm (ASHCA, 2015), while a total of 893,000 total youths lived on farms in the U.S. (NIOSH, 2016). In agriculture, one child fatality occurs approximately every three days (NCCRAHS, 2020; Goldcamp et al., 2004; Hendricks & Hard, 2014), and 38 children are injured on farms daily (ASHCA, 2015). Young workers are 7.8 times more likely to be fatally injured in agriculture compared to the average injury rates of all other industries combined (14.57/100,000 Full Time Employee (FTE) vs. 1.87/100,000 FTE) (NIOSH, 2019).Despite high injury rates in agriculture, there are no fully comprehensive, pre-existing datasets for agricultural hazard analysis and existing datasets are limited in their composition. AgInjury News (Weichlet & Gorucu, 2018), which tracks agricultural injuries by compiling news articles, is one of the most comprehensive existing data sources, but it is limited by the small set of binary variables that it tracks, which do not fully encompass all agricultural risk. VISIMO proposes the development of a customizable decision-support tool, Demeter, which will allow producers to assess and mitigate risk in real time, increasing the safety, efficiency, and productivity of small and mid-size farms. The primary outcome of the Phase II effort will be a prototype of the Demeter application. Demeter will accurately assess risk and provide mitigation suggestions, including cost-benefit analyses, for each suggestion. As development progresses, improvements will be made to the GNN and to the arithmetic formula to improve risk mitigation. Over the period of performance, VISIMO will collect between 5,000 and 7,000 data entries on small and mid-size beef and dairy farms. In addition to the creation of a prototype, Phase II will also result in a dataset for agricultural risks far more detailed and voluminous than any existing dataset. Current datasets rely heavily on limited information identified after a safety incident. In many cases, not all the variables are known because the data source (e.g., a news article) does not include this information. Through the creation of a new dataset, VISIMO will enable new research and education. VISIMO's primary customers include extension agencies, cooperatives, insurers, banks and crediting agencies, professional associations, and educational organizations that seek to address occupational safety issues affecting U.S. agriculture.