Building statistical models to predict the development payment amounts for injury compensation claims have gained some ground in recent years with the advancement of machine learning and AI. While producing a good model that is more accurate than the traditional aggregate triangular actuarial methods remains a challenge for general reserving, there are other applications that cannot use aggregate reserving amounts and need individual claim estimates. Such applications could be producing an early claim estimate for triaging the claim to the most appropriate claim management stream, or, as is the case with this work, assessing the impact of applying road safety initiatives (e.g. road barriers) to the lifetime cost of claims for the claims that happen where these road safety investments were applied. A typical statistical claim estimate model uses information from the claims header – such as injury, hospitalisation, information about the injured person etc.; and the payment history for the claim. In this study we examine how adding unstructured text information with the claim notes impacts the performance of the model. We test different text mining techniques – from simple to more complex, to derive text features. We then use these text features to supplement the traditional claim header and payment data to build our final model. At the end we compare the original and the text enhanced model in terms of accuracy of prediction of the future claim cost, for both claims in their early stages of development, as well as more mature claims.

XAI (Explainable Artificial Intelligence) is a research area focused on developing AI systems that can provide explanations for their decisions and actions. The goal of XAI is to make AI more transparent, interpretable, and trustworthy for end-users. The research in XAI involves developing new techniques and methods to generate explanations for AI models and designing interfaces that allow users to interact with the AI systems and understand the explanations provided. Well-known machine learning methods such as XGBoost predictive classification algorithms, neural network, and support vector machines have been broadly accepted to be helpful in insurance claims management in an academic context. However, due to the limit in complexity and explainability, there is still a long way to go if we want to use those methods practically in the industry. XAI methods provide a gateway for industry users to open the black box and understand how the underlying advanced machine learning method behaves, and why the machine learning algorithms provides such results. Given there are plenty of existing XAI application in claims management, this session will focus on discussing how XAI can be helpful in injuries and disability claims data. Furthermore, this session will provide a framework to evaluate the effectiveness of XAI system explanations. This evaluation framework will allow us to improve the trust with the XAI system, and ground the advanced mechanism in reality.

The reform of the NSW Compulsory Third Party (CTP) insurance scheme was driven by a pressing need to address the significant escalation of small cost claims between 2012 and 2016, which had far-reaching implications for insurance premiums and community affordability. The reform focused primarily on prioritising benefits for the most severely injured individuals and introducing substantial no-fault benefits for those at-fault or with minor injuries while at the same time making insurance premiums more affordable for the community. A critical aspect of the reform was the recognition of the need to manage insurer profits, striking a delicate balance between preventing excessive profits and ensuring fair compensation for underwritten risks. This aim was pursued to establish a balanced and sustainable insurance market that upholds both fairness and affordability for the community at large. This presentation dives into the intricacies of the mechanism devised to manage excessive insurer profits and losses within the NSW CTP scheme. The results from the initial three years of assessments sheds light on the impact of these measures on community premiums. Furthermore, the presentation explores the potential applicability of this mechanism to other insurance schemes in Australia and around the globe.

This session will inform attendees of key developments and challenges recently experienced and on the horizon.

There are broad differences in workers' compensation arrangements in different countries and states, varying by design, benefit structure and claims management. Nevertheless, the mission of workers' compensation systems are often quite similar: to promote safe workplaces, prevent workplace injuries and to improve outcomes for injured workers. In pursuit of this shared mission, workers' compensation systems around the world have gone on their unique journey as they shift towards a greater focus on outcomes and recovery of injured workers. In this presentation, we will compare two workers' compensation arrangements from two different countries and compare how different schemes have applied the worker-centric approach to support workers who have been injured in their line of work. This presentation intends to explore two schemes in particular - WorkSafe Victoria in Australia and WorkSafe BC in Canada. There will be three parts to the presentation: 1) Comparing and examining key differences in both schemes in the areas of scheme design, benefit structure and claims management 2) Utilising a person-centred approach by using (up to 5) personas to show the differences in claims pathway under both schemes - We will explore the claims journey commencing from the time a claim is lodged with the scheme through to the recovery of the worker or exit from the scheme. 3) Deriving a range of metrics to compare both schemes by utilising publicly available data (or scheme data if permission is obtained). These metrics include but are not limited to: - Number of claims - Claim rate, split by physical and mental injuries - Claim rate, split by industry - Average claim size (where available) - Statistics on return to work rates / durations (where available) - Premium rates - Funding arrangements - Payments by benefit type (where available)

This session will inform attendees of key developments and challenges recently experienced and on the horizon.

While insurers and regulators remain focused on funding healthcare that delivers better participant outcomes, consumers increasingly expect healthcare providers to offer services and support that is flexible to their needs. This plenary will examine how insurers, regulators and service providers are evolving the model of treatment and care to uplift the value of healthcare delivery within our injury and disability systems.

Governments are taking steps to make amends for historical practices which are now recognised to have caused significant harm to those involved. These can be seen in royal commissions, parliamentary hearings and inquiries into issues such as governmental policies relating to First Nation Australians and the abuse of children in care. As governments make apologies for these historical injustices, they have also sought to provide justice to those harmed by these practices or failings in the form of redress or reparation schemes. And alongside the financial payments, governments are also providing trauma-informed counselling and other support services, as well as avenues for a direct apology from responsible institutions. One of the challenges for governments with these schemes is the uncertainty in their ultimate costs and hence the funding required to support these schemes. Initial cost estimates are highly uncertain due to the lack of historical data available to estimate the exposed population and an uncertainty in the propensity to apply due to complex circumstances and trauma experienced by the impacted individuals. Participation is also affected by eligibility requirements, societal attitudes, media coverage and the operations of the scheme. This paper will provide an overview of the various redress and reparation schemes around Australia. It will explore how actuarial work within these schemes differ from typical accident compensation schemes and the various unique factors for consideration. The biggest challenge with projection of costs for redress and reparation schemes is estimating the number of potential participants in the scheme. Historical exposure information is often not available or patchy, benchmarking with reference to other comparable schemes may be limited particularly if the redress scheme is unique and the likelihood of impacted individuals applying into the scheme is highly uncertain. Once the number of potential participants is estimated, scheme costs can then be modelled based on the various components of the scheme, such as financial payments, counselling costs and other support services and administration costs. Subsequent to the initial cost projections, governments will need to provide the appropriate disclosures in their financial statements as these schemes become operational, provisions are set aside and updated as experience emerges for these schemes. We will provide insights into unique challenges in considering emerging experience, areas of key uncertainties, including how regulatory changes and societal changes may impact participant numbers.

The AASB 17 standard (AASB2022-9_12-22), titled Amendments to Australian Accounting Standards – Insurance Contracts in the Public Sector and issued on 22nd December 2022 include modifications that apply to public sector entities. AASB 17 is effective from 01 January 2023 for private insurers, with the public sector equivalent to be effective for financial reporting periods beginning on, or after, 01 July 2026. Private insurers’ AASB 17 implementation journey has involved dedicating significant resources and overcoming various challenges to meet the requirements of AASB 17. We would like to explore the below themes for public sector schemes: Guidance on scope of AASB 17 for public sector schemes Expected impact to public sector schemes and differences compared to private insurers in the application of AASB 17 Learnings from private insurers implementation journey and insights from public sector schemes’ early exploration so far Expected end to end impact to systems and processes, for instance, source systems, middle ware, actuarial processes, accounting processes, general ledger and financial reporting Options for AASB 17 solutions

This research paper will cover a data scientists’ approach to using the free-to-use census data at the SA1 level to predict the health condition responses at the SA1 level in the census data. The research aims to identify whether the census data has potential for use in health and life insurance, so that future research can build and extend these ideas further. The paper aims to cover the approach to testing whether there is scope to expand this research with further collaboration from various stakeholders in the industry to gather alternate data sources and increase the power of the population-level census-data. The following topics will be covered: Hypothesis: Can we use census data to predict health conditions? Census data source, data and target Data wrangling census data Data transformations & feature engineering EDA and variable treatment Modelling: Baseline, Supervised and AutoML Future improvements: Other data enrichment, model refinement, alternate target data

Allianz has been a leading national personaly injury insurance provider for over 100 years. Almost all (if not all) innovations that occur within the personal injury ecosystem are social in nature. That is, the implementation of new solutions to complex social problems such as mental health; solutions that ultimately aim to improve the welfare and wellbeing of individuals and communities. Case study: Living well a whole person approach Allianz has long advocated for a biopsychosocial approach to the management of illness and injury. All people are impacted by their personal biological, psychological and social contexts, and when combined with issues related to an individuals’ health, these can ultimately determine the success of that person’s recovery. Many existing health interventions are based on a biomedical model of diagnosis and treatment, the focus is on the specific condition and the therapy evidenced is to resolve the problem. The failure of this model is that it does not consider the psychosocial impacts on a person’s recovery at work. This presentation will explore Allianz’s psychosocial program – Living Well – that was developed in collaboration with The NSW Department of Education (DoE) and Rehab Management. Challenge: The public sector workforce has traditionally had the highest percentage of people with psychological injury claims given the nature of the work they perform, and that remains true today. In 2020-21, 2.7% of the NSW public sector workforce had an active psychological injury claim. Today, TMF agencies account for 18% of all active claims and 46% of all active primary psychological claims in the NSW workers compensation scheme.1 Solution: Through extensive collaboration DoE, Allianz and Rehab Management developed a service tailored to the unique needs of the Department and its employees and provide a program that offers the right psychosocial support following an injury, incident, illness or in other times of need. The program is applied in three settings: Workers compensation claims - Allianz Allianz Case Managers and the DoE Health & Wellbeing team reviewed active claims for the presence of biopsychosocial barriers and suitability for referral into the program. Work-related injury and illness - DoE Trained Department of Education and Allianz case management staff identified individuals displaying psychosocial flags and subsequently referred them for assessment and intervention. Non-work-related cases (Personal Health Conditions) – DoE Living Well has been deployed as an early intervention tool used by workplace managers for situations where a claim or non-work-related injury is not notified, and a need for additional support is identified. Impact The Living Well Program has been operational since December 2020. The outcomes below are based on data as of 20/4/23. Of 354 workers who completed the initial Living Well survey, 61% agreed to participate 172 individuals have completed their full Living Well program. Of these completions; 62% have increased their capacity for work 48% have returned to some kind of work 86% have improved their post-program survey score 67% are primary psych claims Primary Psych RTW improvements: 78% of primary psych cohort had no work capacity at the time of commencing the program, upon completion of the program 69% had increased their capacity for work and of those 73% had returned to some type of employment

The rise of e-scooters has made our cities more accessible but our pathways more hazardous. While the e-scooters status as a vehicle for the purpose of CTP insurance may be doubtful, the potential harm caused by falls from or collisions with e-scooters is not. Building on the local and international research into injury and hospitalisation, we examine how e-scooter injuries might affect CTP schemes. Specifically, we look at what may happen if these are treated as vehicles under various scheme designs, including the interaction with other road users, fault-based compensation, injury severity thresholds, contributory negligence and under-age riders.

One of the most important assumptions is establishing liability estimates and levies/premiums in lifetime care schemes is the inflation to assume over the very long term. Disability support workers and the adjacent industry of aged care workers have and will continue to be in demand. Actuarial assumption related to the future inflation related to these payments have typically incorporated a level of superimposed inflation to capture uncertainties and the potential for DSW wages to increase faster than overall wage growth. In some cases explicit assumptions that attempt to forecast where wages relative to other industries or the economy in total have been made. The actuarial profession has a long history of making long term assumptions where analysis of directly relevant supporting evidence is available. However, economists have built long term models of economic productivity, supply and demand that are arguably better suited to the problem of long run wages forecast. This paper documents a Computational General Equilibrium approach to project national wage growth and disability care wages growth out to the year 2100 incorporating factors such as the rapid growth of NDIS, population aging and technological advances in the care and other industries that improve productivity. The results are presented relative to underlying wage inflation in the medium and long term.

Recent public debate on the National Disability Insurance Scheme (NDIS) has centred on the rising costs of the scheme associated with growing participant numbers but has lacked important context around the benefits this spending provides. The NDIS Act 2013 specifies that reasonable and necessary supports for people with disability should, amongst other things, “develop and support the capacity of people with disability to undertake activities that enable them to participate in the community and in employment”. Hence, the National Disability Insurance Agency (NDIA) has a responsibility to measure how participant funding impacts the achievement of these outcomes. The NDIS outcomes framework was developed to help meet this responsibility. Data from the outcomes framework questionnaires has been collected since 1 July 2016, thus building up a rich longitudinal history spanning seven years. Longitudinal modelling of these data has been used to investigate the link between outcomes and risk factors, including socio-demographic factors, as well as the supports received by participants. The Investment Effectiveness Program (IEP) is a data analytics and policy initiative proposed by the NDIA in early 2022 to further build the evidence around how NDIS funding is related to participant outcomes. The IEP has two broad aims: firstly, to demonstrate the benefits of scheme expenditure at an aggregate level; and secondly, by studying many cohorts and many outcomes, to enable individual participants and planners to derive insights about the funding level and mix which has been effective in similar circumstances to their own. The IEP uses machine learning (ML) techniques to study the effect of NDIS funding on participant outcomes at a cohort level, predicting the increased probability of outcome attainment per dollar expended. ML methods (specifically Gradient Boosted Machines implemented using the XGBoost algorithm) were selected to address the scale and complexity of the project, and to produce model outputs granular enough to be applied in later analytical phases of the project. The longitudinal data collected from the existing outcomes framework is a key input to the IEP. However, the NDIS is expected to benefit the broader Australian economy. Hence, the scope of outcome measurement will need to be widened to include not just participant outcomes but also system outcomes (to measure scheme impact on government service systems and the economy), and community outcomes (to measure scheme impact on community attitudes and society). This talk will present some initial thinking on a new, expanded framework for measuring outcomes impacted by the NDIS, and the relevant sources of data. It will discuss the methodology used by the IEP to demonstrate effectiveness of scheme funding and present some preliminary findings from the analysis.

Spinal Cord Stimulators (SCSs) are surgically implanted neurostimulation devices to treat chronic or persistent pain. In Victoria, injured workers can receive an SCS funded through the Victorian Workers Compensation scheme to treat chronic pain conditions that have not responded to other treatments. However, the medium and long-term outcomes for people with an SCS implant are uncertain. Rapid technological advances have led to exponential growth in clinical trials but with less replication and more uncertainty as to the effectiveness of the devices. Evidence reviews aim to identify, evaluate, and summarise all relevant research findings on a given issue and, importantly, make that evidence accessible to decision-makers. The Institute for Safety, Compensation and Recovery Research (ISCRR) undertook a review examining the current evidence base for using spinal cord stimulators to reduce persistent or chronic pain. The review results found exponential growth in spinal cord stimulation clinical trials and associated reporting. Two Cochrane studies, 21 other systematic reviews, and ten randomised control trials not included in the systematic reviews were identified as having been published since April 2014. Overall, low-level evidence suggests that SCS implants for persistent pain are clinically effective in the short (less than one month) and medium term (up to 6 months). However, evidence is lacking on the long-term outcomes for patients. SCS implants have also shown limited but positive, effects on quality of life, improving disability and functioning, and reducing medication use. The impact of SCS implants on return to work are much less certain, with limited research available. The most recent Cochrane review does not support the use of SCS to manage low back pain outside a clinical trial. The high rate of adverse events reported in Australia, including the large proportion of SCS implants being removed, continues to be of concern. Organisations like WorkSafe Victoria benefit from up-to-date evidence reviews in making decisions about the continuation of funding for spinal cord stimulator implants. Evidence reviews form one part of the evidence-based decision-making process with clinical expertise and patient experiences completing the picture. The best available evidence to date suggests limited support for using spinal cord stimulators to reduce chronic back pain. The evidence review in and of itself cannot answer the question of whether spinal cord stimulators should continue to be used in Australia. Still, it raises concerns and points to the need for more consideration regarding their use.

The Institute for Safety, Compensation and Recovery Research (ISCRR) is a unique 14-year research-government partnership between WorkSafe Victoria (WSV) and Monash University. Focused on improving the lives of Australian workers, ISCRR delivers high-impact research and evidence to guide decision-makers in preventing workplace injuries, promoting worker mental health, recovery from injury and return to work. Through a dynamic and collaborative approach, ISCRR actively engages with WSV and its stakeholders to co-design and implement a research program that prioritises the translation of evidence into actionable decisions and measurable impacts. Operating within a continuous improvement framework, ISCRR remains adaptable and responsive to evolving contexts and priorities. This presentation shares the invaluable learnings derived from over a decade of this research-government partnership. By identifying key success factors in collaborative knowledge creation, we highlight the significance of co-design in the partnership model and proactive monitoring of research impact that has informed decision-making at WSV. Highlights of research and evaluation investments that have informed WorkSafe programs and pilots will be discussed. The ISCRR partnership model serves as a noteworthy example of a successful research collaboration that has led to decision-making within WSV including informing policy, practice, and the development of new initiatives and programs. Its applicability extends to various disciplines and both local and international contexts, making it a valuable framework for other professionals seeking to achieve meaningful outcomes.

Work injury or worker to worker claims have long been a feature of the liability class of business. However, through a combination of the changing nature of the workforce, changes to workers compensation scheme entitlements and the legal environment they have recently been a driver of significant reserve deterioration for liability insurers. Insurers have responded through a combination of premium increase, underwriting standards and higher deductibles for work injury claims. This paper will look at why work injury claims are such a problem for liability insurers, the impact on businesses and whether changes to workers compensation schemes are called for which better reflect the modern nature of employment.