Using Data Analytics to Drive Performance

At the 2019 RIMS Annual Conference a session discussed using data analytics and predictive modeling to assist your risk managment program. The speakers were:

• George Furlong – Senior Vice President, Sedgwick
• Jeff Rycroft – Vice President, Express Employment Professionals
• Zoe Zinn – Workers’ Compensation Manager, Packaging Corporation of America

Predictive modeling is something that gets talked about a lot. Many companies will tout their predictive models. The big question is what are you doing with that data? Too many are just generating reports without using the information to drive change in behaviors and outcomes.

One of the most important elements of data analytics is collecting information from multiple sources. These can include:

• Risk management and claims systems
• Human resource information systems
• Case management data
• Utilization review data
• Medical payment network information
• Pharmacy network information

Predictive modeling can identify patterns that may not be evident to the human eye. It is a tool for the adjusters to assist them in doing their job and driving better outcomes on the claims. Predictive models alone do not produce savings and better outcomes. The adjuster must take action to change the trajectory of the claim in order for savings to be realized.

Artificial intelligence can take predictive analytics and machine learning and take it to the next level. It can learn from the information it gains and suggests what action is needed next. AI is trainable by humans to become more correct with it’s recommendations.

Robotic process automation (bots) works well on repetitive tasks. This is good for processes difficult to custom code and can be used with vendor systems that cannot be modified. The processes can cross multiple systems that cannot be easily integrated.

There are a lot of data projects that fail because they don’t have enough human intervention and the data is not properly organized. Both of these elements are key to success. You also have to approach the project knowing what you are trying to solve for.

As a risk manager the goal of predictive analytics is to identify intervention opportunities that can help you to redirect the claim to the appropriate path. One model that is useful is a large loss model that is used to identify claims that are trending towards adverse development. Another useful model is one that identifies claims with potential for litigation so you can take steps to communicate with the injured worker to help them feel more comfortable with the claims process and prevent litigation.

Data analytics can also assist you in evaluating different vendors in your program including attorneys and physicians. The information collected can provide you with metrics to compare the different outcomes produced.

There is always a story behind the analytics. What is your data telling you? Sometimes you need to peel back the layers of the claim to really identify what is driving costs. One example provided was an injured worker who was recovering poorly. They were able to figure out that the worker was not eating any healthy food and that was impeding their recovery. They used that information to get the employee nutritional counseling to understand the importance of healthy eating.

As a risk manager, push your TPA/carrier/broker to provide you with information that can assist you in making decisions in your risk managment program. All these partners tend to have resources that could be very helpful to your claim managment programs.