Artificial intelligence provides new opportunities to improve occupational safety through data

Finnish Institute of Occupational Health media release 12 October 2023 

The Finnish Institute of Occupational Health completed the four-year research project Improving safety management through data mining – AI Safety with four Finnish industrial companies. The purpose of the project was to develop safety management and occupational safety.

Machine learning methods help analyse large data sets

Organizations are increasingly motivated to use the occupational safety data they accumulate to support safety management. Data is typically available in such large quantities that reviewing it with conventional methods would be extremely burdensome. 

New machine learning methods, such as text mining, that make it possible to analyse large data sets quickly are useful. In addition to conventional occupational safety data, occupational safety can be analysed from typically implicit information related to the operating environment and context. 

The enthusiasm shown towards artificial intelligence in recent years has increased interest in the solutions it can offer to the challenges of business and work life. Meanwhile, the possible applications of artificial intelligence may have become blurred. 

“It is easy to forget that artificial intelligence requires high-quality data in order to provide useful information. With regard to occupational safety data, this means that the data collection and research related to deviations must be systematical and the factors behind human behaviour and the deviation are recorded exhaustively in the text descriptions,” says Head of Research Maria Tiikkaja from the Finnish Institute of Occupational Health.

“The significance of context data is particularly important in preparing forecasting models. For example, an increase in overtime or production volumes may indicate an increase in safety issues in the future. If safety data cannot be linked to other organizational data in the analysis, it is impossible to discover the factors present in the organization’s operations that are associated with the prevalence of safety issues,” says Data Scientist Olli Haavisto from the Finnish Institute of Occupational Health.

New guide to provide tips on developing safety data

A new helpful and useful guide to benefit organizations was prepared as part of the AI Safety project. The guide suggests development methods on how to collect safety data and improve the quality of data. It also provides tips on how and what kind of data organizations need to collect in order to use it more efficiently in machine learning-based development and management of occupational safety.

The data sets allow for preparing for the future developments of occupational safety and taking the necessary measures in a proactive manner. In addition to the quality of data, for example, the information recorded in free-form text, the planning of data systems is important in the context of data collection. The systems need to be user-friendly, easily accessible and facilitate the recording of the required data. 

 “It is a good idea to focus on the quality of data in addition to its quantity. For example, instead of required safety observation reporting quota, you should focus on the factors behind safety in personnel training and how to identify them,” says Maria Tiikkaja.

In an ideal situation, organizations have diverse, analysed benchmark data that combine the data sets collected from various functions of the organization. The results of the analysis provide the required reports for preparing an occupational safety overview and information for developing occupational safety and continuous learning. At best, organizations can plan completely new methods of using analysed occupational safety data to advance the reform of occupational safety management and operations.

About the project

• The project received funding from the Finnish Work Environment Fund and the organizations that participated in it.
• Learn more about developing safety data with the guide (in Finnish): 

  Työturvallisuuden tiedonkeruu – laatua kehittäen: Opas työturvallisuusdatan käytön kehittämiseen organisaatioissa (julkari.fi).

• The Finnish Institute of Occupational Health has also published a non-technical guide to facilitate scientific research on company data and the personal data of employees (in Finnish): 

  Tutkittua tietoa organisaation datasta vastuullisesti ja eettisesti: Opas henkilötunnisteisen organisaatiodatan tutkimuskäyttöön (julkari.fi)

• Browse the project pages: Improving safety management through data mining – AI Safety | Finnish Institute of Occupational Health (ttl.fi)
• Read the final report of the project (in Finnish): 

  Turvallisuusjohtamisen kehittäminen tiedon louhinnalla : AI SAFETY -hankkeen loppuraportti (julkari.fi).