Ai and Workers' Comp Claims Administration Is Not as Disruptive as Advertised

• Publication year: 2020
• Author: Mark Webb Pasadena, California

AI and Workers' Comp Claims Administration Is Not As Disruptive As Advertised

Mark Webb Pasadena, California

Workers' compensation claims administrators are increasingly evaluating and deploying artificial intelligence (AI) and its subsets machine learning (ML) and deep learning (DL) to help reduce claims costs. Generally, in claims administration, AI is designed to accelerate and improve the processing and closing of low-value claims so they do not convert to lost-time claims or, in cases where there is lost time, to keep claims from becoming higher cost PD claims. AI can also assist in medical management by more quickly identifying outcomes relative to approved treatment as well as create analytics by which to measure physician performance.

As the Internet Society stated in its publication Artificial Intelligence and Machine Learning: Policy Paper (November 2017)¹:

Artificial intelligence (AI) traditionally refers to an artificial creation of human-like intelligence that can learn, reason, plan, perceive, or process natural language.

The Internet Society also pointed out that as it relates to ML:

Algorithms are a sequence of instructions used to solve a problem. Algorithms, developed by programmers to instruct computers in new tasks, are the building blocks of the advanced digital world we see today. Computer algorithms organize enormous amounts of data into information and services, based on certain instructions and rules. It's an important concept to understand, because in machine learning, learning algorithms—not computer programmers—create the rules. (Emphasis in original.)

The lexicon of AI can at times be confusing. Consider it, however, in the context of the expert claims examiner. This a person who has had years of experience dealing with both routine and complex claims for benefits. They know QMEs by reputation, applicant attorneys by their tactics, treating physicians by the quality—or lack of same—of their reports, and all the nuances of various WCAB venues. The claims examiner is the person who, in league with all their peers, creates big data (large, complex data sets). It is AI, through user-defined algorithms, that aggregates big data; in so doing, it finds correlations individuals will have neither the time nor the data to uncover. Those insights form rules for claims decisions at critical points in the process.

While there has been considerable discussion about the promises of AI in the workers' compensation context, there is still significant room for deliverables. In a white paper authored by LexisNexis Risk Solutions, "Hype or Reality? The State of Artificial Intelligence and Machine Learning in the Insurance Industry" (December 2019),² a survey of over 300 insurance professionals revealed that technology adopters are primarily using AI and ML to improve fraud detection (70%), triage modeling (64%), and reduce costs (63%). The triage modeling is where many AI service providers stake their claim to value added. In terms of reducing costs, AI can improve economic profiling of physicians in medical provider networks (MPNs) to determine which doctors are not requesting the most beneficial treatments for injured workers (Labor Code §4616.1) or who are potentially abusing requests for independent medical review (IMR).

[Page 9]

In one widely publicized case the potential of AI and its ability to take considerable amounts of data and make it actionable has garnered considerable attention. The Centers for Disease Control and Prevention (CDC), on May 1, 2019, announced:

Researchers from the National Institute for Occupational Safety and Health (NIOSH) together with colleagues from the Ohio Bureau of Workers' Compensation (OHBWC) used auto-coding to determine the cause of 1.2 million workers' compensation claims. The claims were placed into one of three, broad categories: (1) ergonomic related; (2) slips, trips, and falls; and (3) all other categories combined. The number of records

...