Michael Flippo - stock.adobe.com

Analytics in construction has potential to save lives

While minor injuries are on the decline in the construction industry, the rate of fatalities has remained flat, and BI could be a way to bring those numbers down.

Analytics in the construction industry can save lives.

For many organizations, business intelligence is about the bottom line, about finding efficiencies and improving profits. But for construction projects, analytics can help create safe work environments in an industry in which serious injuries and fatalities can be the cost of mistakes.

According to the Bureau of Labor Statistics, there were 1,008 construction deaths in the private sector in 2018, a small spike from the 971 in 2017. Zero may an unattainable number given the nature of construction work, but analytics can be a means toward reducing a number that has held relatively steady while the number of overall construction accidents has dropped.

Dag Yemenu is executive vice president of products at ISNetworld, a software vendor founded in 2000 and based in Dallas that provides analytics tools to contractors working in the private sector. Yemenu recently took time to talk about how the construction industry is using analytics to reduce serious injuries and fatalities.

Among the topics he discussed were the ways most serious injuries and fatalities take place at construction sites and the roles predictive modeling and comparison benchmarks can play in the effort to reduce construction accidents.

In construction, when we're talking about the public sector versus the private sector, is the distinction simply government versus non-government?

Dag YemenuDag Yemenu

Dag Yemenu: In general, the distinction is basically what you mentioned. The public sector would include entities such as utilities that might have state or local administration involvement, whereas when you look at the private sector it's heavy manufacturing, transportation is a big component of that, oil and gas. That would be the main distinction.

You take pipeline companies, refining companies, large auto or paper manufacturing companies, food and beverage companies, and when you think about these industries you need to build warehouses and other facilities, and that's where the heavy construction work comes in. It's not necessarily the government work. It's within the private sector, how you build and maintain assets, and the majority of that work falls under private construction.

In 2017 there were 971 private-sector construction fatalities and in 2018 there were 1,008 -- does that represent a decrease from previous years or do construction analytics show a flat trend going further back?

Yemenu: In general, when we look at safety performance statistics over a long period of time, incident rates have declined. If it's a cut that needed attention or a broken bone, all of that would be reported. However, when we look at the most severe incidents -- what we call serious injuries and fatalities -- they have plateaued. When you look at a 15- or 20-year trend, the smaller incidents have continued to decline but the serious incidents have flattened out. In fact, when you look at [the Occupational Safety and Health Administration's] number of fatalities, there is a slight uptick over the last few years, and that's what we see in the private sector as well.

In terms of trying to understand the injuries, the trend is one thing, but the question of how we explain and understand the precursors and causes is another, and that's where data and analytics come in.
Dag YemenuExecutive vice president of products, ISNetworld

That's why there's a lot of attention being paid to mitigating those serious injuries and fatalities.

Do construction analytics show the way a majority of serious injuries and fatalities take place?

Yemenu: In terms of trying to understand the injuries, the trend is one thing, but the question of how we explain and understand the precursors and causes is another, and that's where data and analytics come in.

There's some data that comes out of OSHA in the U.S., and it gives some insight into what type of incidents are happening. At ISN we have relationships with 650 of the major capital-intensive industries and 75,000 of their contractors, so we have data on incident causes that we collect on a monthly and annual basis. We've looked at the analysis, particularly the serious injuries and fatalities and why they're happening, and we look at the cause and the body parts affected. What we've seen from our data is typically lower extremities and upper extremities -- your legs and hands are by far the areas most affected. In terms of the nature of the incidents that are happening, fractures and dislocations are the top ones, followed by sprains, tears and strains. And when you look at the cause of serious injuries and fatalities, the number one cause is contact with an object or equipment, so being hit by a piece of equipment, followed by falls, slips and trips.

Those are the types of things that, because we have the data, we can pinpoint specific industries or groups of activities -- roofers, or electricians -- and then look at specific causes in that industry.

How deeply has analytics permeated the construction industry?

Yemenu: Analytics has been a growing area from a supply chain and risk management perspective -- that's kind of the broad umbrella of applications. It has evolved over the last five years, primarily trying to find the specific causes of injuries and illnesses. I would say that it has increased, and part of it is that there is more data available and more tools available to not only do trending and descriptive analytics but predictive analytics, and that's what we need to get in front of these incidents. We need to try to correlate what is causing these serious injuries.

And what's the key construction data that's being input into analytics models and dashboards to try to reduce serious injuries and fatalities?

Yemenu: It all starts with good data. The majority of the construction work is being done by contractors -- not employees of the oil companies or the manufacturing companies -- and from that perspective, previous safety performance data is universally one of the main data points that's being collected. If you are a contractor providing services to a client, the client is going to ask for at least three years of your safety performance data. What types of programs and procedures you have is another data point that needs to be looked at. And then, typically, major organizations that hire third-party contractors also have their own data that they collect observationally by walking around the job site and seeing what types of infractions there are, or what near misses are happening. Another one is incident data, because if I have a job site and there's a near miss I need to understand what happened and some of the controls that failed.

A lot of data is being collected from observational audits as well as the incident management systems.

How can analytics be used to increase the safety of construction workers -- what is the leap from the dashboard to the job site?

Yemenu: There's the dashboard that gives you a good picture of the risk profile of a company, their past performance -- whether they have had a fatality when providing service -- and then there's an audit score so I can have a number that tells me whether I can use this company with reasonable risk or whether I'm taking too much risk by having work awarded to this contractor. That's data-driven decision-making in terms of understanding the risk of hiring Company A vs. Company B.

There's the use of data from a trending perspective. If I'm a food manufacturer with 120 soft drink-producing plants across the country, I want to pinpoint -- using data -- which of my locations are lagging behind in terms of performance and which ones are ahead. That way I can direct my resources and my effort in those areas and locations and facilities that seem to have some challenges versus others. You can't have infinite resources, and analytics help clients pinpoint specific locations, specific plants, specific companies that seem to have challenges.

Another use of analytics is, from a benchmarking perspective, looking at how good is good. If I'm an auto manufacturer, what are my peers seeing in terms of incidents in terms of body parts that are being injured, and how do I benchmark myself against them? That way I can direct my resources to certain areas.

What about preventative measures?

Yemenu: Predictive analytics is now coming into the picture, which is really the next phase of not only understanding the story that we have with the data, but being able to predict. We're lucky to have data from 75,000 companies, and we apply predictive models to discover that if you are a construction worker, which attributes of a construction company are correlated with better or worse safety performance, and through that correlation analysis we can infer that if a company has a certain profile, it has a higher risk of incidents. It helps us look forward instead of just at past data. Organizations are trying to stay ahead of these risks.

As far as the technology itself, are there analytics platforms designed specifically for the construction industry or are contractors using Microsoft Power BI or Tableau or other business intelligence software?

Yemenu: It depends on the size of the company. When you look at construction, you obviously have the large, sophisticated construction firms that have the tools to look at their performance data, incident data and field observation data. We see and hear about good old Excel still being used. Tableau is becoming a widely used tool, and we use Tableau in our operations. Power BI for a little bit more sophisticated organizations. But when you look at the statistics for contracting companies, the majority of them are small- to medium-sized companies. A good majority of them are five-, 10-, 20-, 30-person companies and they are working and giving services in a high-risk industry. These organizations don't have the sophistication to use [analytics] tools, so organizations like us, which have data from 75,000 contractors, provide dashboards and benchmark data for them. If I'm a roofing contractor or electrical contractor, I want to know how I'm performing compared to my peers, and we provide those easy-to-use tools to help them understand and navigate because not everyone is sophisticated enough to have their own Tableau or Power BI dashboard.

Is there resistance to analytics adoption in the construction industry or are most people open to it?

Yemenu: Most people are open to analytics. We continue to get more requests, and in the industry we see more appetite for data, particularly benchmarking data. That tells us that the overall sentiment is changing. Companies want to know where they stand. Risk management and performance benchmarking are now becoming C-suite level discussions, and owner and founder-level discussion, because clients that are hiring contractors are looking at the same data. If you are not meeting certain safety performance benchmarks, you may not get a bid. The culture has evolved to where analytics is an executive-level discussion. Resistance hasn't been a challenge. In fact, on the opposite side, we're seeing more appetite for data.

Is there anything else you'd like to add about the role of analytics in the construction industry?

Yemenu: In general, applying risk management in capital-intensive industries continues to grow and be data-driven. We're seeing organizations that may be less mature in their risk management capabilities -- still using paper documents for collecting data -- have moved over to better discovery tools, better data collection. And we see the continued growth of data not only in contractor management but overall risk-management application, and benchmarking is becoming key. We're excited about the trend. Obviously there's a lot more to investigate.

We're excited about machine learning applications. The safety and risk-management world still uses paper documents and there's a lot of information on paper, so we've started using machine learning to extract data out of, for example, OSHA logs. Trying to capture that data from PDF documents and trying to make sense out of them is becoming important, and machine learning and analytics tools are helping the industry in that area.

Editor's note: This Q&A has been edited for clarity and conciseness.

Dig Deeper on Data science and analytics

Data Management
Content Management