University collaborates on healthcare research into worker injury
In collaboration with Chicago and Denver-based Peers Health, an organization aiming to reform workforce case management and help workers get back to their jobs after illness or injury, the University of Michigan is currently working on a two-year research project with the goal of optimizing health care plans and return-to-work times for employees on injury or sick leave.
Currently, organizations like Peers Health are finding strong relationships between a worker’s ability to perform their job’s duties and their overall well-being and health. By improving this healthcare infrastructure, the University and Peers Health are hoping to increase productivity in hospitals, insurance companies, offices and more.
According to Brian Denton, an Industrial and Operations Engineering professor, who is one of the principal investigators of the project along with Jenna Wiens, assistant professor of computer science and engineering, the research will be applying several research methods, such as algorithmic investigations and statistical tests, to a comprehensive data set of anonymous workers’ compensation cases.
“What we want to do from that data is try to learn what works well and what doesn’t work well using data science approaches, so algorithmic approaches, for learning what decisions are likely to lead to somebody being able to get back to work in a safe and timely manner,” Denton said.
The research will use machine learning, along with statistical and industrial operations methods, to create a model for how a patient can be treated for injury or illness and when they can return to work.
Rackham student Haozhu Wang, who is leading the student research group collaborating on the project, said the wealth of data the project has to work from will benefit the project immensely.
“This data set that we are going to work with is really kind of novel,” Wang said. “It’s a huge data set which contains the information of more than one million patients … To my knowledge, this is one of the the largest data sets by far (to be analyzed by machine learning).”
Wang also noted this is some of the first research using machine reinforcement learning, which investigates data sets through trial-and-error as a human often would, that looks at occupational healthcare.
Peers Health CEO Jon Seymour said the focus on return-to-work data is relatively new in occupational health data analysis as well.
“The focus of the data set is return-to-work, which is specifically how to get patients back to work from injuries and illnesses,” Seymour said. “It’s an important facet of medicine that is often underappreciated … Getting someone back to work from an injury or illness is equivalent to returning them to their productive endeavors.”
Seymour said the decision to work with University researchers was based on the advanced technological power, such as Mcity and the highest amount of research and development spending out of all public U.S. universities, and knowledge available.
“Here, these technologies, like artificial intelligence, are so cutting edge that the more we looked at it, the more we felt, to get the most out of it, we needed an academic component,” Seymour said. “These technologies are the ones that will eventually parallel a self-driving car, (they will) be able to actually interpret data and develop a policy around what to do in a given situation, and make proactive recommendations to doctors and other stakeholders.”
Denton reiterated Seymour’s self-driving vehicles example, saying the technology and research in artificial intelligence can apply to the research and data procedure the University and Peers will be working on with worker injuries through creating the learning models that rely on artificial intelligence to synthesize the data.
“Self-driving vehicles are trying to learn over time to become more and more effective using data they collect from multiple sources, and in a way that’s exactly what we’re trying to do here with the data that Peers is providing — many different sources of data that we bring together to build learning algorithms to improve upon decision making,” Denton said.
Seymour acknowledged there is considerable commercial interest going into this research — Peers Health develops and sells guidelines in the return-to-work industry — and he said he believes the research has the potential to go even farther into the health care field.
“The implications of what we’ll learn here go beyond the niche of occupational health care and really extend broadly to general health care, and potentially every medical episode that goes on all day, every day,” Seymour said.