Reimagining Occupational Health and Safety Compliance with AI

6 mins read

Over 6.2 billion hours of worker productivity and a gross sum of $163.9 billion are lost to work-related injuries every year. Can Artificial Intelligence help cut these losses by augmenting workplace occupational health and safety compliance monitoring? Let’s discuss. “It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts.” – Sir Arthur Conan Doyle, Sherlock Holmes. The importance of data cannot be emphasized enough when it comes to decision-making. However, today industries are generating noise at an alarming rate. Up to 78% of all data manually collected, processed, and stored by your company could be simply “useless”. Shocked? In our previous article, we have discussed in detail how manual data processing could bottleneck your HSE journey to Goal Zero and strongly recommend you go through it before proceeding further. The focus of this article is to discuss how AI uses the same data “smartly” to provide the conventionally unobtainable granularity of strategic and actionable insights to your HSE team.

Table of Contents

AI Preventing Safety Incidents

Ensuring safety compliance is the duty of on-ground HSE officers. A ratio of 1:200 (safety officers to workers) for the usual running of factories/refineries and 1:50 for high-risk operations like construction has been stipulated as ideal by IOSH, but it is not uncommon for this figure to touch even 1:500. To add to that, the area of supervision under a single safety officer can be as high as 34000 square feet!

CCTV cameras are of some assistance but going through long hours of footage is a fatiguingly repetitive activity capable of restricting the scope of human observation.
Not wearing a glove while operating a machine is a high risk, which can lead to injuries and can easily go unnoticed

A single HSE officer must keep their eyes peeled for compliance violations across all the activities that are going on in their zone of supervision. In exactly 100 such activities, they can expect 19 safety violations and 2.8 incidents just waiting to happen on average. This is where the concept of YOLO comes in.

YOLO – You only look once

AI can observe and report PPE non-compliance issues in real-time, enabling the reviewer to take corrective action before something goes wrong. AI powered solutions continuously monitor visual or sensor-based sources of data at your workplace for indications of compromise. Depending on its training, AI can detect a wide range of operational risks to processes, assets and people. You ONLY HAVE TO LOOK ONCE!

Amputation risk imminent - another instance of safety compliance violation that can easily go unnoticed

Why YOLO?

Monitoring the workforce for HSE non-compliance is a monotonous part of an HSE officer’s job. They are additionally responsible for several other tasks like planning and reviewing policies, advising trainers, preparing safety education, conducting risk assessments, identifying bottlenecks, preparing safety plans, etc., which demand their own, much more justified share of time and attention.

AI Responding to Safety Incidents

Despite everything we do, accidents are bound to happen. But how we respond to such situations determines everything from the potential damage to repercussions and even repetition of similar mistakes in future.

An ideal accident response involves assessing the damage, securing the scene, and reporting. But contemporary siloed approaches to HSE hamper planning and coordination, and with it, the ability of the administration to respond to incidents efficiently.

For proper planning and coordination, visibility across the entire chain of command is imperative. This is where the unity of data and interface comes into play. AI dashboards can unite all relevant data and automate processes generating insights, ensuring they are available to all important stakeholders in the hour of need.

Then, every incident is an opportunity to learn something new. RCA (Root Cause Analysis) is the process of determining what went wrong and contemporary RCA methods, just like contemporary incident responses, are bogged down by a lack of detailed endpoint data, process logging, and event correlation.

“RCA tools need to be used in conjunction with an understanding of context to make correlations between events and draw insights for corrective actions.”

How good is AI with RCA? Turns out, AI can help with all 3 stages of RCA – detection, analysis and remediation. It can help eliminate blind spots, provide better prioritization context and implement long term solutions, not just workarounds.

This can be answered better considering the question you have been waiting for, all this while.

Can you trust AI for Safety? – A Story of learning from Mistakes

How good is AI with RCA? Turns out, AI can help with all 3 stages of RCA – detection, analysis and remediation. It can help eliminate blind spots, provide better prioritization context and implement long term solutions, not just workarounds.

“AI is better at accomplishing tasks that are repetitive in nature.”
Because of its imitating abilities, AI has the quality to identify informational patterns that optimize trends relevant to the job. In the case of humans, the informational pattern is uneven. When there is what we call a “human error”, determining the true nature of it can be nightmarishly difficult.
While AI also commits mistakes, it is consistent with them. It is going to repeat the same mistake over and over until eventual discovery and remediation. The AI will then never commit that mistake again.
How does that translate to practicality? Let’s assume an unattended machine sparks fire at one of your locations. You can rest assured that all your other locations are now safe from a similar incident – AI is going to report as soon as it spots an unattended machine of similar type.

Statistically, the deployment of AI proved to be a game changer for a petrochemical enterprise in the APAC region. Within a month into deployment of T-Pulse (an outstanding HSE AI), dangerous workforce exposures went down by 80%, the number of non-compliance issues reduced by 10X and a gain of 8% was observed in overall productivity.

Summing Up

Usually, organizations are constantly reminded that health and safety failure is expensive. Of course, it is, but that is just one side of the coin. Pragmatically, the benefits are much greater. A Comcare paper reveals an average return of $5.81 for every dollar invested by companies into workplace HSE programs.

Despite that, with the global market getting increasingly competitive, HSE is one of the worst hit departments. The global HSE budget was cut by over 13% between 2019 and 2022 and further cuts are still planned. The staff numbers too have been reduced by 22% since 2010.

In such extraordinary circumstances, AI is the very elixir of life for an industry under pressure of increasing production with tighter deadlines, lower HSE budgets, and where the goal – of zero incidents, remains an ever-moving target.

Share this blog post via