Project Story: Preventing Hand Injuries in Manufacturing

About the Project

Workplace safety has always been a critical yet often overlooked aspect of industrial operations. What inspired us was the persistent pattern of hand-related injuries in the manufacturing sector, especially given that many companies run repeated campaigns with limited success. We aimed to break that cycle by making prevention smarter—not just more frequent.

What Inspired Us

While reviewing OSHA’s Severe Injury Reports, we noticed that hand injuries were consistently high across states and sectors. Many of these could have been prevented with simple interventions. This insight sparked our mission: turn raw data into actionable prevention strategies.

What We Learned

How to work with messy, real-world data and clean it using SQL, Excel, and Python
The power of Large Language Models (LLMs) like ChatGPT in parsing narratives to detect root causes
Interpreting chi-square analysis to identify statistically significant contributors
The importance of pairing capital investment with human training for long-term impact

How We Built It

Data Cleaning & Filtering: We cleaned OSHA injury records to focus on hand-related incidents in manufacturing using NAICS codes.
AI + NLP Categorization: Using LLMs, we categorized each injury narrative into:
- Root cause
- Prevention step
- Intervention type (Capital, Human, Both)
Statistical Analysis: Applied chi-square tests to identify the most injury-prone sub-industries and factors.
Visualization: Built interactive dashboards highlighting injury trends, severity levels, and actionable recommendations.

Challenges We Faced

Inconsistent Data Formatting: The injury descriptions varied widely in clarity and length, which made categorization tricky.
Interpreting Narratives: LLMs required fine-tuning and validation to ensure accuracy in assigning causes and interventions.
Aligning Insights with Action: Turning data into meaningful and usable recommendations for real-world scenarios took deep collaboration and iteration.

Final Outcome

Our project offers a comprehensive, AI-powered roadmap to reduce hand injuries in manufacturing. By combining data science, domain research, and thoughtful strategy, we empower companies to move from reactive responses to data-driven prevention.