Artificial Intelligence and Real-world Implications in Disability Management

 

“Leveraging Artificial Intelligence (AI) in Personal Injury and Disability Management.”  I’ve delivered presentations on this topic at the International Forum on Disability Management (IFDM)  in Vancouver (BC), the Personal Injury Education Foundation (PIEF) Conference in Perth (Western Australia) and to audiences at online webinars and roundtables. While I could attend only a fraction of seventy-five or so sessions and hundred other presenters, what I’ve learned through my attendance and interaction with presenters, providers, and delegates (including injured workers) at these events has implications for everyone working in this field.  Here are a few highlights related to AI in personal injury and disability management.

AI is everywhere …even if we are reluctant to admit it

Microsoft quotes an International Data Corporation (IDC) study finding 70%+ of Fortune 500 companies have now use CoPilot in their organizations (Ignite 2024: Why nearly 70% of the Fortune 500 now use Microsoft 365 Copilot November 20, 2024,  Microsoft Hong Kong).

About 75% of knowledge workers worldwide are using AI in their work (Microsoft and LinkedIn, 2024 Work Trend Index Annual Report, May 8, 2024 https://news.microsoft.com/2024/05/08/microsoft-and-linkedin-release-the-2024-work-trend-index-on-the-state-of-ai-at-work). 

Use of AI by university staff ranges from 62% for sessional staff to 81% for senior staff with academic staff at 75%   (McDonald, P., Hay, S., Cathcart, A. & Feldman, A. (2024). Apostles, Agnostics and Atheists: Engagement with Generative AI by Australian University Staff. Brisbane: QUT Centre for Decent Work and Industry. https://eprints.qut.edu.au/252079)

What I learned in presenting this information is that AI use among professionals in personal injury and disability management is in its initial stages. This is not to say personal injury and disability management (PI&DM) professionals are unaware of AI’s potential. About a third of delegates I encountered work employers who have implemented enterprise-wide AI solution such as CoPilot or ChatGPT enterprise editions. What has been missing is the direct and specific information they need to effectively use AI tools for their specific tasks. Other delegates report overly restrictive prohibitions against using AI at work, although many admit to simply using AI on their own smartphones or laptops or on home computers. This last point is consistent with the Microsoft/LinkedIn study that found 52% of people who use AI at work are not waiting for their employers to catch up, are bringing using AI on their own devices, and are reluctant to admit to using it for their most important tasks.

AI is pervasive in the PI&DM research and provider community

Although some organizations are restraining or restricting application of AI, the research community is applying the technology to significant effect. One presentation at the IFDM demonstrated this effectively. “Using Ensemble Random Forest Algorithms to Predict and Determine Return to Work Intervention and Pathways” (Mohamad Amrizad Bin Ruslin & Nabilah Binti Ahmad, presentation at IFDM 2024, Vancouver) might not have the catchiest title but the results were exceptionally noteworthy.

Working for Malaysian Social Security Organization (SOCSO) – PERKESO, the research focused on supporting workers with disabilities through personalized Return to Work (RTW) interventions. The challenge is familiar to every workers’ compensation and disability insurance program administration: to optimize case management to reduce the duration and cost of rehabilitation while increasing success rates. The research employs a collection of AI tools to predict and recommend tailored RTW interventions based on individual worker data (e.g., injury type, job demands, rehabilitation needs) at or near the claim acceptance. The results demonstrated high predictive accuracy, increased efficiency in allocating resources, and faster return to work. The researchers also introduced their new  work on creating their own “official disability guide” based solely on Malaysian disability cases. This makes sense; while quantifying “impairment” is standards-based assessment, “disability” duration and impact are dependent on external factors including access to care, employment law,  and societal acceptance.

Use of AI among providers to the personal injury prevention, worker’ compensation, and disability insurance industries is advancing quickly. Advanced analytics such as those demonstrated by Clara Optics provide risk identification using AI’s assessment of unstructured data sources was a good example. Individual claim alerts display when there are changes in sentiment, pain, medication, or psychological impact.

Michelle Barratt for Arriba Rehab Management (RM) presented their propensity model (Lean-On_Learning_Assistance-LOLA). This up-front AI application accurately predicts worker risks, allowing RM to tailor service delivery pathways in line with evidence-based protocols, resulting in improved RTW outcomes, reduced case costs, and durations.

Amanda Johnston demonstrated the integration of AI into case KINNECT’s CareLever platform of claims management systems. The Dashboard  in their “Manage” module provides case-specific, real-time status of outcomes and quality, client centricity, and even file “hygiene.”  One interesting feature was the digitization of psychosocial questionnaire instruments such as the ARIBA (Assessment of Risk for Interpersonal Violence or Abuse), DAS (Depression Anxiety Stress Scales) and Fear Avoidance Belief Questionnaires (FABQ): The Fear-Avoidance Belief Questionnaire. This method avoids the delay between the time a case manager determines the need for such assessments and the return of scored results by allowing for delivery to the client by SMS messaging, automated scoring and return to the case manager.

Several presenters spoke about their implementations of AI include CoPilot. Presenters and  service providers in the exhibition space demonstrated the power of leveraging AI to create promotional and educational materials. For example, a complex regulatory-change news release transforms into podcast discussing the changes through a generative AI application; that’s a particularly powerful demonstration of making information more accessible for certain learning styles. One service provider now creates podcasts from curated RSS (really simple syndication) feeds for their clients.

Functional assessments are always a challenge in PI&DM. In countries such as Australia, where long distances may be involved, the application of XTRA’s AI application to perform virtual measurements such as range of motion from real-time video consultation sessions was impressive.

This power of AI to eliminate delays in the sequence of case management events was evident several presentations and product demonstrations I viewed at both the IFDM and PIEF conferences. PI&DM professionals can help clients achieve the best outcomes when diagnostics, assessments, and treatments occur without delay.

Compliance with prescriptions and medical instructions can also lessen duration of disability. Several presenters and providers demonstrated AI systems helping clients more actively participate in their treatment and rehabilitation through SMS messaging and notifications  to prompting medication use,  exercises, symptom reporting, and appointment attendance.

Dr. John McMahon (Navigator Group) presented his practical application using machine learning in PI&DM. One application was Jin, an avatar-based virtual claims examiner. While not perfect, the AI-driven Jin interacts with clients, collects data, and can do so in more than one hundred languages. The power and potential of this technology illustrates the breadth of where AI is taking us.

Implications for PI&DM

These recent conferences and interactions with audiences on AI illustrate the challenges and opportunities AI presents PI&DM. It is evident that there is an explosion of work in the AI field but AI literacy among professionals is lagging. A recent study found that more than 70% of university staff and instructors were using AI in their work, but gaining their knowledge about AI from friends, family, and YouTube videos. Informally, the students and PI&DM professionals I encounter typically have experimented with AI but lack fundamental understandings about this technology, its ethical uses, and limitations.

Leadership in PI&DM organizations must make AI strategies and guidelines a priority. They must also recognize that this is not a “one and done” exercise. Whatever you decide, you will have to revisit your strategy often.

Administrators have a tough task when it comes to AI. They must adapt to the changing environment, select applications and products, while maintaining privacy, confidentiality, and regulatory responsibilities. Even these tasks require raising AI literacy levels throughout the organization. They must recognize that restricting AI use is a futile effort. With 80% of  knowledge workers already using AI, enabling ethical, regulatorily compliant use is the only option. Gaining and sustaining professional levels of AI literacy needed to achieve that will be an ongoing challenge in terms of time, effort, and cost.

Individual practitioners need a professional level of AI literacy in order to critically assess the AI tools they use and to understand both the upstream use and downstream consequences of AI proliferation. 

For educators, have similar challenges professionally. Raising and maintaining our own AI literacy while incorporating AI into our courses to better prepare PI&DM professionals is an immediate priority.

Final thoughts

A 2023 IBM publication concluded “AI won’t replace people, but people who us AI will replace people who don’t.”  To put this more directly for our profession, AI will not replace PI&DM professionals, but PI&DM professionals who use AI will replace those who don’t.

There is no avoiding what AI is bringing. Only increased AI literacy among PI&DM professionals, administrators, leaders will optimize the impact of AI for the clients we serve.