How to break through the Flattening EHS curve with Computer Vision AI?

Picture this: I’m sweating it out on the cross trainer with my headphones on, tuning in to Carla’s latest episode of The Accidental Safety Pro Podcast on HSI. Carla and I used to work together at Shell, so I know she’s a powerhouse when it comes to safety. And let me tell you, this episode did not disappoint.

As I listened intently, Carla mentioned something that really struck a chord with me. She spoke about how the EHS performance has become stagnant – stuck, even – and I couldn’t help but nod my head in agreement. The dataset she referenced was fascinating.

According to the Occupational Safety and Health Administration (OSHA) in the US and the Health and Safety Executive (HS&E) in the UK, overall EHS statistics have plateaued in the past decade. While the number of incidents has decreased, the rate of decline has slowed down. For instance, in the US, the total recordable incident rate (TRIR) has reduced from 5.0 in 2003 to 2.8 in 2018, but the rate of decline has been much slower in recent years. See the figures below extracted from NSC. 

While these figures represent a global aggregate, the same trend is evident at the micro-organizational level. During my time at Intenseye, I have collaborated with multinational corporations across various industries, and without exception, they all face a similar challenge to ‘plateauing’ EHS performance.

This is not new to me, as I already encountered similar findings in my Master’s Thesis in 2007. The figure below is an extract from that work, and it shows that the conclusion I reached back then still holds true today. 

“… HSE has passed through three major eras; namely, compliance with HSE standards, development of the HSE management system, and cultivation of an HSE culture on and off the field or site. Companies initially considered HSE as a set of mandatory standards imposed by the government. Upon realisation that compliance with standards alone does not provide a sustainable solution to incident prevention, companies developed a structured system to comply with HSE standards effectively and efficiently. Contrary to initial expectations, even after putting the standards in place by means of a structured methodology, the incidents did not stop; although there was a dramatic decrease in the number of incidents. In order to further reduce the number of incidents, the concept of ‘HSE Culture’ was invoked. ‘HSE Culture’ lies deeper within the organization and it is difficult to observe.”

In a podcast, Adam Grant and John Amaechi put it well when they said that “Culture is defined by the worst behavior tolerated”. This concept applies to safety culture as well and we must gain a better understanding of human behavior to be able to assess and improve the EHS performance. Although organizations have continued to implement several new approaches to improve the stagnating EHS performance (e.g., training programs, safety audits, behavior-based safety programs, and enhancements in technical safety management systems), and while these initiatives have proven to be effective to some extent, they have not been able to break the plateau in the journey to zero harm.

Why has the curve flattened out? 

Based on my quick survey, here are the typical reasons of this plateau formation: 

• Organizations have focused on compliance with regulations and standards, but this has led to a culture of compliance rather than a culture of safety. 
• EHS initiatives have been reactive, addressing incidents after they occur rather than being proactive and preventing them. 
• Workforce is becoming more diverse, and language and cultural barriers can hinder effective communication and training. 
• Work environment is getting more complex and sophisticated which requires a lot of interaction between employees, machinery, or robots.

My colleague David Lemon, one of my thought partners at work, argued that the reason for the flattening out of EHS performance may be due to the fact that the measurement i.e. TRIR, is statistically invalid from the outset, referring to a recent paper published on the topic, based on 17 years of data collected over 3.2 trillion worker hours. One of the main conclusions of this study is “In nearly every practical circumstance, it is statistically invalid to use TRIR to compare companies, business units, projects or teams”. David also added “without a good measurement, we will always face this flattening issue, which is well-known and documented, but unfortunately, measuring safety remains a ‘holy grail’.” This sentiment takes me back to the title of my final dissertation presentation, i.e. ‘In Search of the Holy Grail.’ 

Given these insights, two pertinent questions arise: Is an accurate measurement of EHS performance still an unattainable goal, and is it still challenging to observe and understand human behavior and organizational culture in the 2020s?

While traditional EHS initiatives have had some success, adopting technology and transitioning to a data-driven approach is essential to step up the game. Computer Vision AI technology offers a solution to help achieve this objective, by detecting potential hazards and preventing incidents in real-time, thereby promoting a safe work environment. Here are some ways in which computer vision AI can help:

• See the potential hazards 24/7 in real-time, which might otherwise go unnoticed at job sites. This empowers a data-driven decision-making process that prioritizes high-risk activities or hazards and raises alerts in real-time.
• Shift the focus from lagging to leading indicators by analyzing the data generated by AI systems to identify potential issues before they occur and take steps to mitigate them.
• Identify and optimize precautionary actions in real-time. Analyzing data from AI systems will disclose certain patterns of behavior and work practices that may contribute to incidents or injuries. This information can then be used to develop targeted training programs that help workers improve their skills and work more safely.
• Harness the power of humans and enable the implementation of human performance principles in the workplace. By providing workers with instant feedback and alerts, AI systems encourage the frontline workers to raise their concerns and contribute to finding solutions to the problems that they face.
• Finally, it is crucial to generate precise and reliable quantitative datasets that serve as a “source of truth,” facilitating consistent assessment and comparison of EHS performance across different departments, sites, companies, and industries. These data points can help companies establish more dependable and efficient key performance indicators at both the individual and team levels.

In conclusion, the implementation of computer vision AI technology can offer a powerful solution for organizations seeking to create a safer workplace, while breaking the plateau in route to zero harm. By utilizing its ability to provide a more accurate and reliable understanding of workplace behavior, and aligning it with human performance principles, organizations can foster a culture of safety and accountability that benefits everyone. Taking advantage of this technology can help benchmark EHS performance more consistently and achieve the shared goal of zero harm. 

Join the growing number of organizations embracing AI and make a real difference in creating a safer future for all! Schedule a demo with our team.