By Mike Bishop, VP of Product and Technology
Even without realizing we are doing so, we make predictions all the time. How much time is it going to take to get to work this morning? How will my friends react to my dinner cancellation? Which team is going to win the World Cup this year? What is the best way to handle a workers’ compensation claim?
With more experience, we are often able to make better predictions. Yet, even with this knowledge, humans typically have a larger margin of error than data-driven computer models and algorithms (as we can clearly see by my failed World Cup bracket). Although we may evaluate similar data elements as machines, we are prone to human biases that ultimately skew our predictions. Awareness of these biases may help improve our ability to make predictions, as can practice, but large amounts of data and subconscious biases can still affect our forecasts.
Particularly in the workers’ compensation industry, stakeholders deal with an immense amount of information. In just one claim, we may have to account for an injured worker’s physician visits, pharmacy fills, ancillary needs, financial needs and mental wellbeing, among countless other variables. Although the human brain has the ability to store massive amounts of information, it can fall short of tracking all moving pieces and connecting all variables together.
This is where technology has the ability to propel our subpar predictions to excellence and ultimately improve outcomes.
Unsurprisingly, predictive models are more accurate than human guesses. These models can process a large amount of data without bias and look for patterns that you or I may have missed.
The words “machine learning”, “artificial intelligence” and “predictive analytics” are three important terms to understand as we explore the application of technology to workers’ compensation. Machine learning means giving computers the ability to learn without being explicitly programmed; artificial intelligence is the display of this learning. Predictive analytics uses machine learning, among other techniques, to process data and create hypotheses of what could happen based on historical data.
Predictive analytics provides us robust insights in a practical and applicable way. We can utilize this technology in claims to predict what may happen next, based upon the claimant’s demographics as well as medical activities and injury (e.g. prescriptions filled, treatment, fractures versus sprains). This technology gives adjusters greater insight into the possible trajectory of a claim and can identify which injured workers might be most at risk.
However, there are two important takeaways from the relationships between predictive analytics and human expertise.
The first is that humans are incredibly important both in the building and interpreting of the models. Computers are able to process vast amounts of data, but can often fall into the trap of mistaking what are valid inputs as well as cause and effect relationships. For instance, a computer might very well accept the logic that someone with more birthdays is likely to live longer than someone with fewer birthdays. Take a second to process that sentence. You can easily see that, although funny, that is not really a valid insight, since having more birthdays does not cause you to live longer. The human brain is able to distinguish this and program predictive models to do so as well, but without that nuanced understanding, the models are not nearly as effective. Experts in the industry need to work hand-in-hand with the data scientists involved to build the models with appropriate factors.
The second important takeaway about predictive analytics is that it does not tell us what will happen, but rather what may happen and with what level of certainty.
For instance, a predictive model may have chosen Spain to win the World Cup based upon various factors, as another model may predict a negative outcome in a claim. Yet, what the model predicts may not be the actual outcome of a claim. Why is this?
As we discussed in our previous article, these technologies need the expertise of the humans involved in workers’ compensation claims. Armed with the knowledge of objective predictions, humans can shift the projected outcome of a claim. Technology cannot make the correct decisions alone (despite how ubiquitous and “intelligent” it is becoming); those who know the claim inside and out and work in the field every day are the ones who are able to do so.
It is true that predictive analytics provides stakeholders a better platform to make quality decisions; the informational and predictive burden is lifted when these technologies are used properly. In turn, this allows adjusters to focus on appropriate levels of management and oversight to help injured workers continue on the best path to recovery. However, the decisions of adjusters, nurse case managers, physicians and pharmacists are just as important as the predicted outcome the model provides. These complementary and interwoven abilities are the key to better decisions.
Although it’s often difficult to acknowledge the shortcomings of our own brains, the ability to do so allows us to consider the exciting possibilities at our fingertips. The combination of powerful predictive models and our ability to turn the information from those models into positive actions will undoubtedly improve claims outcomes and the overall workers’ compensation industry.
Humans and technology have complementary skills. It is time we start using that to our advantage.
About Mike Bishop
Mike Bishop is the Vice President of Product and Technology for Mitchell Pharmacy Solutions. He leads the Technical Product Development organization and is responsible for technical product strategy, direction and execution. Bishop has been with Mitchell for 12 years holding various leadership roles in Product Delivery, Operations and Software Development. For more than twenty years, Bishop has worked in Enterprise Software Development with experience ranging from small startups to Fortune 500 companies including Oracle and Siebel.
Bishop holds a Master’s degree in Computer Science from the University of Illinois and an undergraduate degree in Computer Science from Georgia Tech.
About Mitchell ScriptAdvisor
Mitchell ScriptAdvisor, the PBM solution built for the workers’ compensation industry, leverages technology and expertise to connect the ENTIRE claim. From a fully customizable portal experience to industry leading networks and comprehensive clinical management, ScriptAdvisor provides effective, efficient, and successful management of your claims. ScriptAdvisor delivers visibility beyond an individual prescription. It integrates with bill review and managed care solutions for a more holistic approach to claim management. ScriptAdvisor combines all of this with robust analytics and outstanding customer service to empower better outcomes for all. Read more of Mitchell’s thought leadership at mitchell.com/mpower.
Disclosure:
Mitchell is a WorkCompWire ad partner.
This is not a paid placement.