Skanska USA, Skanska Sweden, Webcor Builders, Obayashi and concrete contractor Lithko Contracting have signed on with the Predictive Analytics Strategic Council, which was launched earlier this year to share and analyze safety data from each firm in the group.
The council relies on the artificial intelligence-based system from image-analysis firm Smartvid.io, dubbed “Vinnie,” that aggregates information from each of the firms to help predict future incidents and hazards. The new companies join founding members Suffolk, Barton Malow, DPR Construction, JE Dunn, Messer Construction Co., Mortenson, Shawmut Design and Construction and the Bouygues Group, as well as insurance broker Aon.
The group is building on the results of a Smartvid.io and Suffolk study showing that Vinnie learned from Suffolk’s data to predict roughly one in five safety incidents with 81% accuracy.
The council’s firms have agreed to share digital project data from a period of more than 10 years to build predictive models for safety and risk. By working together in this first-ever type of industry-wide collaboration, member companies are able to identify trends they might not have otherwise seen working alone, said Jit Kee Chin, Suffolk executive vice president and chief data officer and chairperson of the council.
“No one company has quite the volume of data it takes to create really good artificial intelligence,” she said. “The council provides a way to get together to explore some of the potential that advanced analytics has to offer.”
The five new companies will bring more data to the ultimate goal of increasing worker safety, according to council member Marni Hogen, director of health and safety at Mortenson.
“It’s exciting because the more information we have, the better our insight will be,” she said.
Hazard prediction
The council members, who are contributing their safety data anonymously, meet quarterly to discuss the operational, social and business impacts of predicting safety hazards, according to Josh Kanner, CEO of Smartvid.io and technical chair of the council.
The council, which convened for the third time last week, has made progress over the past few months, Kanner said, by collecting data to train deeper predictive models and writing a white paper on the challenges and best practices to bringing predictive analytics data to executives and project teams.
“The successful implementation of these technologies and techniques are as much process as they are pure technology,” said Kanner, who added that the white paper will be published early next year.
Hogen said to date Mortenson has contributed information about its projects that include location, size, dollar value, duration and incident information. In the future, council members are considering providing even more data such as manpower, RFIs and work hours.
While some of those data points might not seem important, Hogen said every bit of information that is plugged into Vinnie is useful.