CompTalk: What Is Deep Learning?

Mike Paczolt of Milliman discusses the benefits of deep learning and AI in workers' comp claims management.
By: | June 18, 2021

The data and information available in a workers’ comp claim are vast and, at times, complicated. In the workers’ comp space, inconsistency within the claim is a common issue.

Now, we have the capabilities of artificial intelligence (AI) and deep learning to not only solve these complications of inconsistency, but to improve the current operations of workers’ comp claim managers.

Mike Paczolt, principal and consulting actuary, Milliman, introduced the proficiencies of deep learning and how it can serve as a valuable tool to workers’ comp professionals during his National Workers’ Compensation and Disability Conference (National Comp) CompTalk, now available on-demand.

Paczolt stressed that the goal of deep learning was to make the jobs of workers’ comp claims managers easier and to automate mundane tasks that are both time consuming and repetitive. He also dispelled a few myths about what deep learning and AI is not: self-aware, magical or a one-size fits-all solution.

Rather, deep learning and AI shines in “solving complex problems with large amounts of data and distilling it into something much more efficient,” Paczolt said.

While AI is machine learning that finds data correlation by utilizing statistics, deep learning is more complex in its abilities. It attempts to mimic neuropathways of the human brain, which allows it to solve more difficult problems with less human input.

A type of deep learning is natural language processing (NLP) that computes data from text excerpts or passages to accomplish a variety of automation processes. NLP can be utilized for knowledge extraction and transcribing and summarizing large excerpts of text, all of which are time saving techniques.

Pazcolt said the deep learning and AI marketplace is constantly evolving, which will result in more changes and innovations as the technologies are further explored.

The role that AI can play in workers’ comp is invaluable. What used to be a manual process managing large data claims has now become easier, and in some cases, more accurate. Some of AI’s capabilities in the workers’ comp space include claim triage, fraud prevention and sentiment analysis.

In terms of the future of AI and deep learning, Pazcolt stressed that while trends are pointing toward an increase in automation, it doesn’t mean automation will be replacing previous capabilities. Rather, the future of these technologies will continue to automate those mundane tasks in a more frequent manner. Other capabilities, like speech dictation, wearable and improved algorithms, are other aspects Pazcolt expected to improve over time.

Pazcolt also hoped for an improved explanation of both AI and deep learning for workers’ comp claims professionals. With a better understanding of both entities, claims managers will better embrace the technologies.

You can view the full session on-demand here. Check out more CompTalks on the National Comp website. &

National Comp — the National Workers’ Compensation and Disability Conference — is back! We’re planning an in-person show for October, 20-22, 2021 and we’re excited to see everyone while still adhering to all safety protocols set forth by local and national health authorities at the time of the event.  Register today!

This year, we’ll feature seven tracks — from core content on medical and pharmaceutical management, claims and return-to-work, plus new and expanded avenues to explore like risk finance and injury prevention. All of our educational sessions are chosen for their ability to deliver sound takeaways and ideas that attendees can use right now.

In the meantime, National Comp will continue bringing you free virtual, educational content through our digital sessions series and our CompTalks program. Register today to make sure you don’t miss a digital session and check out our on-demand CompTalks library.  Missed a session? Watch it here on-demand.

Emma Brenner is a staff writer with Risk & Insurance. She can be reached at [email protected].

More from Risk & Insurance