RiskVision for Safety Excellence Group

Inspiration

The alarming frequency of hand injuries in the manufacturing industry inspired RiskVision. Recognizing the critical need for enhanced safety protocols, our team for the Safety Excellence Group embarked on a mission to leverage data-driven insights to prevent these injuries. We aimed to create a solution that identifies the root causes of hand injuries and provides actionable recommendations to mitigate risks and ensure a safer workplace for all.

What it does

RiskVision analyzes incident descriptions from manufacturing facilities to identify common causes, sources, and scenarios of hand injuries. It provides data-driven insights and visualizations that highlight high-risk machinery, frequent injury mechanisms, and workplace conditions contributing to these injuries. The analysis offers actionable recommendations to improve safety protocols, enhance worker training, and implement protective measures. Although RiskVision is not yet a tool, it is in the pre-production stage and has the potential to be developed into a comprehensive tool in the future.

How we built it

Data Preparation & Cleaning : Isolate incidents related to hand injuries and narrow down by searching keywords in relevant columns (like "Final Narrative").
Keyword Extraction : Using NLP techniques like TF-IDF and N-grams, we extracted common terms and phrases related to hand injuries.
Topic Modeling : We employed Latent Dirichlet Allocation (LDA) to group common scenarios of hand injuries and identify patterns.
Action Verb Analysis : We analyzed action verbs to understand how injuries occurred and identify high-risk activities.
Co-occurrence Analysis : We identified combinations of machinery and injury types from the narratives to pinpoint frequent injury scenarios.
Data Visualization : We created various charts and graphs, including bar charts, pie charts, stacked bar charts, heatmaps, and line charts, to visually represent our findings with matplotlib.
Presentation : We compiled our insights into a comprehensive powerpoint presentation and have included our analysis outputs into word docs to effectively communicate our findings and recommendations.

Challenges we ran into

Data Quality : Ensuring the accuracy and completeness of incident descriptions was crucial for reliable analysis. We had to clean and standardize the data to remove inconsistencies and extra spaces.
Complexity of NLP Techniques : Implementing advanced NLP techniques like LDA and TF-IDF required a deep understanding of the algorithms and careful tuning of parameters.

Accomplishments that we're proud of

Successfully extracting and analyzing key terms and phrases related to hand injuries using NLP techniques.
Identifying common injury scenarios and high-risk machinery through topic modeling and co-occurrence analysis.

What we learned

Machinery such as presses, conveyors, and cutting tools are major contributors to these injuries.
Inadequate machine guarding and insufficient worker training significantly increase the risk of severe injuries.
Additionally, we discovered the importance of protective gloves and the impact of workplace conditions on injury severity.

What's next for RiskVision for Safety Excellence Group

Expand Data Sources : Incorporate more incident descriptions from diverse manufacturing sectors to enhance the robustness of our analysis.
Refine NLP Techniques : Continuously improve our NLP models to better capture and analyze injury narratives.
Enhance Visualizations : Develop more sophisticated and interactive visualizations to provide deeper insights.
Collaborate with Industry Partners : Work closely with manufacturing companies to implement our recommendations and monitor their impact on workplace safety.