By Cliff Belliveau, myMatrixx Chief Innovation Officer and Phil Walls, Chief Clinical Officer
So far, in 2023 we’ve seen Artificial Intelligence (AI) rapidly become a major topic with massive implications for nearly every facet of life — including in the fields of healthcare and workers’ compensation. The ongoing development and startling proficiency of Machine Learning (ML), Natural Language Processing (NLP) and Large Language Models (LLMs) in applications such as ChatGPT are causing leaders and experts to rethink the immediate and long-term futures of everything, from patient care to drug therapy to customer service. In part one of this article, we discussed some of the most significant opportunities for AI in workers’ compensation, including virtual assistants, fraud and waste reduction, workflow automation, medical data analysis and better outcome predictions.
While there is every reason to believe the AI future can be bright, it doesn’t mean it will be implemented smoothly and perfectly. As with many new technologies and developments, opinions on AI tend to fall roughly into camps of optimists or pessimists. While optimists focus on the potential upsides of AI that we discussed previously, pessimists tend to focus on the risks and dangers.
Instead of one camp or the other, the best position to take for AI may be one of vigilant optimism. To ensure the industry benefits from AI, workers’ compensation leaders and stakeholders should understand and take a clear look at the challenges and obstacles that this technology presents, both in general and specific to our industry. However, while they should be taken seriously, these are also hurdles that can be overcome with proper planning.
In this second part of Navigating our AI Future in Workers’ Compensation, my colleague Phil Walls is joining me to help untangle some common objections and misconceptions about AI. With a background in AI, I’m someone who is far from a pessimist — but I have certainly encountered many common arguments against AI. That’s why I’m going to present five of the most relevant objections to AI in our industry and have Phil respond with solutions and ways to proactively address these challenges.
1. AI is just a fad
It seems like nearly every year a new buzzword takes over the technology sector and finds its way to our industry. Any recent discussions and proposals around adopting blockchain technology in workers’ compensation during the peak of cryptocurrency speculation are a good example of this phenomenon. Although there are practical reasons, which we don’t have time to cover here, why blockchain in workers’ compensation likely won’t materialize anytime soon, it illustrates the type of conversations that are quick to spring up around new technologies.
When discussing the future of AI, it’s natural to think this thought and have this type of concern. After all, AI seems like the hot new thing now, but like other trends, it can be easy to wonder if it will be replaced in the cultural conversation by something else within a year, whether it’s virtual reality or call centers on the moon.
Phil Walls: Although technological fads and buzzwords certainly come and go, there have also been significant innovations that were initially dismissed as fads. When the internet was first discussed in the early nineties, many compared it to the CB radio craze of the seventies and eighties. In addition, few people predicted what an enormous impact smartphones would have on the world when Apple launched the iPhone in 2007.
There are multiple indicators that AI is more comparable to the internet as a major shift that is here to stay. In fact, it is already present in many forms, from virtual chats to analyzing financial markets. In drug research and development, pharmaceutical companies have been using machine learning for years to discover new treatments and make breakthroughs.
Barring some unforeseen obstacle, AI is set to continue creating dramatic efficiencies and innovation across every area of health care. Companies that ignore it and don’t have a strategy run a considerable risk of being left behind.
2. AI will put workers out of a job
This actually might be the most frequent single objection we come across at conferences or events covering this topic. Basically: “Talk of automation sounds great, but won’t it put everyone out of work?” To take it a step further, if smart machines automate every industry, would there even be a need for workers’ compensation?
Pessimistic scenarios for AI and the future of work range from people becoming lazy to widespread unemployment, to something out of the Terminator movies in the worst cases. Once again, complete automation is a long way off, but today it would be perfectly reasonable for a leader in workers’ compensation and risk management to ask if implementing AI means supporting something that could eventually put them and their workforce out of business.
Phil Walls: The short answer: every new technology that promises automation has come with this warning and yet people are still working today. The most famous historical example was probably the 19th century group of English textile workers known as the Luddites, who saw timesaving and laborsaving innovations such as automated looms as a threat to their livelihood. In response, they would conduct clandestine raids to destroy or sabotage modern textile factories, which is why today the term Luddite is used for people who are skeptical of and resist technological progress.
While no one can predict the future, past trends point to humans being essential to the workforce for the near future. While advancements in AI have been remarkable and are likely to continue at a rapid pace, machine learning is still only applicable in highly specific circumstances and it does not have the adaptability and general problem-solving abilities that people have. Additionally, while pessimists tend to focus on the jobs that technology and automation take away, there are also new jobs and opportunities that they create that are often harder to predict.
What AI really has the opportunity to do, especially in the near-term, is help people do their jobs better and more efficiently. With workers’ compensation facing a labor shortage and a significant part or our workforce entering an enormous retirement bubble, most claims professionals in workers’ compensation will be more likely to tell you they need all the help they can get.
3. Who supervises AI? Oversight and quality assurance
While this question has been common in science fiction for decades, it’s become one that we’ve had to start taking more seriously in recent years. The more we have intelligent computers running things — including making decisions, designing systems and implementing changes — the less familiar human beings will be with these processes.
This means that if an AI makes a decision that causes problems or is even harmful to people, by the time we notice it, it may be too late to prevent significant damage. What’s worse, because the AI has been in charge of the system, we may not even know how to fix it or turn it off.
Phil Walls: This topic should be taken seriously when implementing AI, or any new system for that matter. Looking back in history, oversight and quality assurance to minimize human error have been major challenges, so we should look at AI with a similar lens. This means that any system involving AI should be tested carefully and designed with multiple fail-safes.
This also points to another reason people won’t be out of work anytime soon. As we stand today, AI applications require continuous training by humans. While the speed and scale of an AI process will be able to far outpace a person, there will still need to be someone looking over their proverbial shoulder. In workers’ compensation, this can and should mean ongoing oversight from real people including data scientists, programmers, physicians, pharmacists, nurses and claims professionals for years to come.
4. Data is too segmented and fragmented for AI to make accurate predictions in workers’ compensation
When it comes to analyzing and interpreting workers’ compensation data, AI can only be as good as the data it has to work with. If an AI algorithm or learning model is tasked with identifying health risks and making predictions about outcomes, the predictions will be less accurate if it only has access to a small set of data. Likewise, if AI is able to analyze a much larger, feature-rich sample, we can be more confident in the predictions it makes.
In workers’ compensation, companies from third-party administrators (TPAs) to pharmacy benefit managers (PBMs) often only have their own data to work with, which could lead to partial pictures or distortions in any analysis or prediction an AI makes. Data interoperability in workers’ compensation is a topic I have discussed previously in terms of traditional data analysis, and is one that will need to be addressed if we want to truly harness the predictive powers of machine learning.
Phil Walls: This is a legitimate challenge and requires our industry to move forward with creating secure standards for data interoperability and sharing. Although health data is a highly sensitive and private set of information, being able to access a population’s or a patient’s complete medical picture can lead to better outcomes and more streamlined treatment costs. Two prominent examples of industries that manage sensitive information and still develop functional data interoperability are telecommunications and finance — think about being able to use your ATM card at a competing bank or making a call to another cell network for an idea of how and why this is important.
Even as general health care is working to move forward with sharing secure health data, there are several reasons why it’s a bigger undertaking for workers’ compensation. First and foremost are the multitude of state-based regulations and systems that can affect how health data for injured workers is created and administered. However, without overcoming this fragmentation of health data, the ability of AI will be limited in our industry, at least when it comes to identifying risks, predicting health outcomes and potentially lowering treatment costs.
5. Can workers’ compensation close the innovation gap?
Some fields just don’t adapt to innovation as quickly as others do, and workers’ compensation is definitely a sector with an innovation gap. As with adopting data interoperability standards, the variety of laws, policies and regulations that exist among state workers’ compensation systems can have a lot to do with it. An organization handling claims for multiple states has to have complex and rigid systems in place that are costly and time-consuming to upgrade.
Additionally, insurance companies of any type tend to be risk averse for obvious reasons. Businesses whose very livelihood is to calculate and minimize risk are typically not going to be early adopters of an unproven technology. For these and other reasons, it’s not hard to see workers’ compensation continuing to hold an innovation gap when it comes to putting AI in place.
Phil Walls: The short answer to this question is – yes, and AI offers a unique opportunity to actually solve the problems that have kept workers’ compensation behind the innovation curve. For example, an AI could be trained to become a compliance expert that can create integrated workflows for companies that function across state lines. Moreover, even the most traditionalist analysts and actuaries may find themselves adopting AI-driven risk assessments once the results are tested and verified. However, it is important to note that regulations, whether they are promulgated by the various divisions of workers’ compensation, boards of medicine or boards of pharmacy, must also keep current with this rapidly developing technology or we may not realize its full potential.
To an even larger degree than adapting to information technology and the internet, the organizations that are able to embrace innovation, including AI and machine learning, will most likely find themselves at an extreme competitive advantage in the coming years.
There are challenges and obstacles to implementing AI and ML in workers’ compensation, but the potential upsides are substantial. Any organization involved in caring for injured workers should be working to develop a forward-looking strategy to carefully integrate this promising technology for the ultimate benefit of workers and stakeholders in our industry.
About Cliff Belliveau
Cliff Belliveau is the Chief Innovation Officer for myMatrixx. He is responsible for leading the initiatives that enable automation, advanced data analytics and visualizations for internal and external stakeholders. Belliveau is an accomplished information technology leader with a track record of delivering solutions to complex business problems. He has over 25 years of experience in all facets of information technology – from managing high performance teams to systems architecture, application integration, software development, operations and support. He is also the inventor of two awarded patents.
He is a lifelong resident of Tampa and enjoys spending time with family, water sports, traveling and coaching youth athletics.
About Phil Walls
Phil Walls is the Chief Clinical Officer for myMatrixx. He joined the company in 2006 and oversees all aspects of myMatrixx’s clinical program including drug utilization review, formulary management, drug regimen reviews and targeted intervention with prescribing physicians.
Phil is a clinical pharmacist with over 40 years of experience in pharmacy, healthcare informatics and workers’ compensation. Previously he served in leadership positions within the industry with Health Information Design, Inc., PMSI and Cigna Healthcare of Florida, Inc. He is a published author and frequent national speaker on clinical issues in workers’ compensation.
In recognition of his contributions to the industry, Phil was honored to be named CompPharma’s 2015 Person of the Year and to receive the Dorland Health People Pharmacist Award. Phil received his Bachelor of Science in Pharmacy from Mercer University School of Pharmacy.
About myMatrixx
myMatrixx, an Evernorth company, offers best-in-class pharmacy services for workers’ compensation programs that include: formulary and network management, utilization management, claims processing, home delivery and specialty pharmacy care and physician outreach programs.
Working with the financial and risk management leaders of organizations, myMatrixx helps reduce the pharmacy cost associated with injured workers through innovative programs, business analytics and robust clinical protocols and expertise.
myMatrixx is committed to providing predictable pharmacy benefits management to get injured patients back to work and their lives safely.
Disclosure:
myMatrixx is a WorkCompWire ad partner.
This is NOT a paid placement.