Correspondence model of occupational accidents

Area of Science:

• Occupational Health and Safety
• Data Science
• Risk Management

Background:

• Workplace accidents pose significant challenges to workforce productivity and safety.
• Existing models for accident prediction often lack a generalized framework applicable across diverse industries.
• A comprehensive analysis of accident data is crucial for developing effective preventative strategies.

Purpose of the Study:

• To develop a generalized model for the diagnosis and prediction of accidents within the Spanish workforce.
• To establish a risk-injury pattern for early-stage risk assessment and decision-making in companies.
• To create a framework for the automatic management of work-related accidents.

Main Methods:

• Utilized observational data from 7,519,732 accidents across Spanish companies over eleven years.
• Classified accidents into a novel 19x19 risk-injury contingency table.
• Employed correspondence analysis to identify a three-axis structure representing distinct risk and injury groups.

Main Results:

• The correspondence analysis revealed three distinct groups, forming a general Spanish pattern for risk and injury relationships.
• Identified frequent and likely associations between specific risks and injuries, aiding in early risk assessment.
• The model's structure provides phenomenologically understandable characteristics for each risk-injury group.

Conclusions:

• The developed model offers a generalized approach to workplace accident prediction and management applicable internationally.
• Facilitates early-stage risk assessment and decision-making by highlighting common risk-injury relationships.
• Highlights the need for a standardized classification framework for risks and injuries to enable cross-country comparisons.