Nicholas Bernini is manager of predictive analytics for Pittsburgh-based Predictive Solutions Corp., which makes safety management software and performs consulting to help predict possible safety lapses. The company says its goal is zero workplace fatalities in the not-too-distant future. Bernini, who has a B.S. in mathematics and an M.A. in computational mathematics, was a data scientist for several different companies before joining Predictive Solutions. He recently spoke to ENR Boston Correspondent Johanna Knapschaefer about how data improves workplace safety in construction.

Where does the data for the analyses performed by come from?

When we do our analysis, the two pieces of info we need to capture are the injuries incidents that we’re trying to predict and safety observation data points performed by managers, foreman or even the workforce. We also look at manhours worked on a specific site, training of personnel, production schedules, video monitoring and surveillance, images, and any sensor info such as from forklifts or wearable gear.


We also like to tap into the data of people from the HR [Human Resources] system, but we do this on an anonymous basis looking at trends rather than people’s info.

Will knowing where workplace danger and hazards are more likely to occur eliminate the need for some craft safety training and on-site safety supervision?

No, we are not trying to replace any jobs, but we are really giving the individuals in charge of safety a way to focus their efforts today. We’re telling them where their risks of injury are going to happen. We will have to focus the BBS [behavioral-based safety] on areas of high-risk based on a predictive model. Many times companies already have limited resources, so this [predictive analytics] is allowing them to take those limited resources and focus them in the right areas.

What construction companies or others are using predictive analytics to enhance safety now?

While I am not at liberty to name individual clients, our industries include construction, manufacturing, oil and gas, the military, and the state, local and federal government, ranging in size from small businesses to the top five in the global market. Our demand is definitely going up based on my workload and a majority of our work is construction.

Give us a concrete example of how it will work and what it will cost?

Cost is extremely variable for our highly customizable services. Typically if a client is enrolled in our predictive analytics, they are streaming their observation data points to us in real time while our models continuously monitor that info, looking for areas of risk. Once a risk is identified, typically we alert the client through an email or text or other notification method saying this location is at high risk over X number of days, here’s what’s causing the problem, and here’s a solution on how to mitigate it. The client can then log in for more detail.

What could a company of 100 or fewer employees do to use these predictive analytics methods? 

Using a tool called SafetyNet, a Web and mobile-based software for electronic safety observation collection, a client can go out with their mobile phone, collect their observations and sync the data back to the database in the Cloud, and we can maintain it for them. If you are a small client of 100 or fewer, we can pool a bunch of clients together to make that data large enough to work. We have clients ranging from enormous construction and manufacturing companies to small mom and pop specialty construction companies, so we can aggregate data from smaller clients with data from the large construction companies.