Analyzing and Predicting the Influence of Weather on Health, Safety, and Environment in an Operational Setting

A growing body of anecdotal and published evidence points to job site exposure to dangerous weather conditions as an important contributor to worker compensation claims. As the climate changes and the frequency of extreme weather events continues to increase at an alarming pace, the number of weather-related workplace injuries will likely also rise. Acknowledging the negative impacts of workplace hazards on employee well-being and productivity, many of the world’s largest companiesprioritize the establishment of a safe workplace environment. Despite the demonstrated success of artificial intelligence (AI) in transforming numerous areas of business, AI has not yet found widespread adoption as a tool to help mitigate workplace hazards. Additionally, while there exists a robust and regulated set of protocols governing workplace safety, businesses have been slow to take advantage of the growing corpus of data, including weather data, available for machine learning-based approaches. This work presents a solution developed at IBM that aims to leverage this data and the latest AI-based techniques to greatly reduce time-to-value in the extraction of occupational health and safety (OHS) insights from data.

The solution, Health and Safety Insights Advisor (HSIA), can scan and process hundreds of documents relating to both process and occupational hazards, along with incident records and weather data in seconds. It does so in response to natural language queries that allow a human to quickly arrive at conclusions about actions that ultimately lead to an optimally productive work environment while minimizing safety risks. This outcome is empowered by training a model that maps machine-derived features (including factors such as temperature extremes) against historical incidents. Beyond analyzing historical risk, HSIA can ingest updated records that improve AI model performance. This presentation will highlight some of the weather-related OHS insights derived from the tool. Furthermore, practical applications of the lessons derived from the analysis are demonstrated in a weather insights dashboard that allows users to set weather thresholds learned from HSIA.