Huge data that may be wilting away in company servers could have a bigger purpose in the organization than just satisfying compliance requirements. Experts are saying it may just be the key to the age-old quest for total injury prevention.
Through data analytics (also called, predictive analytics), some companies are finding good use for data gathered over a number of years in bringing injuries and fatalities down — which can only be good for the bottom line. Anything from safety-related stats (safety inspections, near-miss reports, lost-time injuries) to broader corporate data (employee demographic, equipment maintenance reports, performance bonuses), data experts say all have the potential to produce great intelligence for safety applications.
This is what Vancouver-based mining firm Goldcorp has realized following a six-month pilot project in one of its mines in Ontario that put its vast corporate data to the test using Deloitte’s data analytics software.
“We looked at five years of data, collected over two million data points, and we started to see this data is telling us something,” says Paul Farrow, senior vice-president, people and safety for Goldcorp. “In some cases, (it) reinforced what we already knew but now we actually have quantitative, statistically significant data to work with; in other cases, it kind of showed us some areas which we were not expecting to see.”
Goldcorp took various data from the Ontario mine — including safety incidents, production, times, production bonuses, employee socio-demographic profiles, climate measurements, equipment maintenance and geo-spatial data — and fed them into the analytics program developed by Deloitte.
The system then produced intelligent reporting on trends related to health and safety. In a nutshell, the data analytics provides information that allows Goldcorp to identify key drivers of safety incidents. For instance, Farrow says, through the analytics the company found certain types of employees have a higher propensity to a workplace injury.
“We found the work propensity for an individual to have an incident is much higher if you are single, versus married,” says Farrow. “Married with children even further decreases (the propensity).”
Other key drivers of incidents Goldcorp has found have to do with production and bonuses. The company found the propensity for safety-related incidents increases with bonuses. However, with the most experienced workers — and therefore the highest bonus earners — Goldcorp sees the reverse is true, as the propensity for safety-related incidents actually decreases with this group of workers.
The data analytics also found, overall, January, March, July and September are higher-risk months for incidents. There is an increase in incidents among aboveground employees between September and March, while underground crew experience the highest number of incidents in July and September.
On a weekly basis, higher incidents occur during Tuesdays and Wednesdays. Goldcorp also found other key drivers for incidents, including weather, hours of daylight and other environmental factors; relationship between aboveground and underground workers and their propensity for injury; and the influence of demographic factors (marital status, experience) on incident occurrence.
While getting this kind of intelligence from corporate data is valuable for organizations, what they do with those analytics is key to really achieving the safety performance improvements they desire, according to Dave Stewart, chief operating officer for Deloitte Analytics based in Toronto. Organizations that are harnessing analytics capabilities in an effort to solve business problems tend to be more successful than those that don’t understand how to leverage their data, Stewart suggests. It’s an innovative way of looking at old problems to generate new insights in reducing and mitigating risks to people and companies.
“Data in its smallest form is just data,” says Stewart. “Data that’s been analyzed creates some form of information; information combined with expertise creates intelligence… and that may help create predictive forecasts.”
He says Goldcorp understands this.
“At the end of the day, the data is giving us the direction,” says Stewart. “Whether some companies believe to take that direction is really (about) culture, the ability to adapt to change.”
Farrow says the greatest value in data analytics will be derived on the frontlines.
“The site supervisors and managers now have that information, they can see it on a graph. It’s really driving a behaviour change,” Farrow says. “We are now getting at the precursors of what is causing these incidents to happen. The leading indicators can help to make sure that anything you put in place — whether it’s additional training or more visible leadership in the field — the leading indicators will have to start telling you how well you are actually doing that and (if there is) a correlation with improved safety.”
Incident predictions Multinational construction firm Lafarge has been evolving its big data utilization to improve its safety performance over the last 10 years. With about 80,000 employees across the globe, Lafarge has a more focused approach to data analytics, using two leading indicators: employee engagement by management team — under its Visible Felt Leadership initiative — and job observations.
“Imagine all these observations going on all over the world, all these conversations, all these interactions, all these engagements with managers and employees, and employees and employers,” says Chris Roach, director of health and safety with Lafarge Canada in Calgary. “We’re trying to capture the details of those conversations to give us some solid data of where we should focus our attention.”
These leading indicators are fed into a custom software program which produces trending reports that identify what types of risks exist at every site, the percentage of safe versus unsafe acts at various sites, and which sites are more likely to have a safety-related incident in the near future, explains Roach. With the analytics tool, Lafarge is able to identify the top risks or sites that need special focus.
“For example, we see five per cent of unsafe acts are related to working at heights, then we zoom in on working at heights, zoom in on fall protection training, zoom in maybe on re-issuing our working at heights standard in that area,” says Roach. “It allows us to get ahead of those things before the accident happens. We’re just observing the unsafe acts and we’re able to capture them and then act on them.”
According to its 2011 global sustainability report, Lafarge has dramatically decreased its global lost-time injury rates from 8.35 in 2002 to 0.63 in 2011, thanks in part to the company’s data analytics initiative. The same positive improvements are being achieved by other organizations that have started embracing the potential of data analytics for injury and fatality prevention.
Denver-based Cummins Rocky Mountain (CRM), a distributor of Cummins and Cummins Power Generation products, started implementing data analytics in 2009. CRM uses SafetyNet, a safety management system software from Oakland, Pa.-based Predictive Solutions. Within 12 months, the company’s recordable incident rate was reduced by 76 per cent and lost-time injuries decreased by 88 per cent. The company enters data collected through safety observations and audits (leading indicators) into the SafetyNet system.
The system then generates a report that predicts where the next incident may occur, based on leading indicator data, explains Griffin Schultz, general manager for Predictive Solutions. The predictive model behind the SafetyNet technology was developed through a collaborative study with The Language Technologies Institute at Carnegie Mellon University (CMU) in Pittsburgh.
The study used leading and lagging indicators from actual workplace data across 250 work sites. The result was an accuracy rate of between 80 and 97 per cent in predicting incidents. The Language Technologies Institute is the same CMU department that helped IBM develop the Watson supercomputer that is now helping doctors diagnose rare diseases.
SafetyNet generates “red flag” reports which tells a company which of its work sites have a higher risk of safety incidents, says Schultz.
“As you collect more safety observations — leading indicator data — you fuel the data analytics and the predictive models get better and start predicting where your incidents are going to occur, and then you can prevent them,” he says.
Tim Smith, safety manager for Cummins Rocky Mountain, talks about the value of data analytics in reducing the company’s injury rates.
“It enables us to establish a relationship between leading and lagging indicators so we can understand whether we’re looking in the right operations and aspects of our business operations to identify opportunities to get better,” says Smith in a video posted on Predictive Solutions’ website.
What the future holds Still in its infancy, data analytics for injury prevention takes many forms and goes by different names — predictive analytics, leading indicators — and the technology that fuels this capability will need a series of enhancements to maximize the potential in predicting and preventing the next injury.
For Deloitte, helping companies understand and take advantage of their vast corporate data is key.
“What we are looking at is what other points of data do these companies have access to that they have traditionally not tapped into,” says Stewart. “We’re trying to understand what we don’t know… does that data add value when they combine it with other data sets?”
One thing experts are certain about is that leading indicators are overtaking lagging indicators in providing good intelligence for improved safety and preventing injuries and fatalities in the workplace.
In mining, industry observers note that while the number of lost-time injuries has consistently gone down over the last decade, the number of fatalities seem to have plateaued — and safety advocates continue to look for new and better approaches to prevention.
“Traditional approaches to date have not seemed to reduce the percentage of fatalities,” Stewart says. “So there’s a drive to take a different approach to the same challenge and the difference here will be looking at all of these data sources and converting them into some form of intelligence.”
The advantage and value of leading indicators also becomes more prevalent as companies start to improve their safety performance.
“I do think (data analytics) is the next level of health and safety,” says Lafarge’s Roach. “We’re still on that path of finding better ways to capture the data, finding more consistent ways to capture it. You get to a certain level of health and safety and you start to see your lagging indicators levelling off.
“We’re at that point now where our incident rate is quite low and we’ve got to do some innovative things to try to get it down to world-class.”