The March 2021 Future of Health Seminar delved into the health system’s most pressing problems and explored how healthcare providers, governments and insurers are using data to create better patient outcomes and improve resource efficiency.
Trigger Warning: This article discusses suicide ideation, discourse and mental health. For confidential help, call Lifeline on 13 11 14 or beyondblue on 1300 224 636
The annual gathering – this year held entirely online because of COVID-19 – brought together healthcare experts, insurers and actuaries from around the world, including Australia, the Netherlands, and the United States, as well as government policymakers and industry advisers, to shed light on some of these issues and how they are being managed.
The overarching theme running through the seminar was the growing role of data science, analytics, and technology in healthcare – how data is used by healthcare providers, governments, and insurers to create better patient outcomes and improve resource efficiency. The seminar also discussed some of the emerging challenges as interactive technology in healthcare becomes more prevalent.
To understand the central role data has in such a system, it’s worth taking a brief look at changes to healthcare across the world.
Healthcare costs are rising faster than economic growth across all OECD countries, creating new pressures on existing traditional healthcare systems. Changes to aged care, chronic disease management, new technologies and fragmented primary and acute health systems are contributing to healthcare costs, says Anna Stevens of Ernst and Young.
This is forcing governments to examine new ways of providing healthcare. Multiple speakers noted that the healthcare system is shifting from a volume-based approach – providing the full gamut of services to everyone – to value-driven care, which examines whether a particular patient achieves a particular health outcome.
Stuart Houston, Director of Financial Services for Google Cloud, sees a huge focus among both health insurers and healthcare providers to make consumers more central within the healthcare ecosystem, rather than merely an end recipient.
Health data can be a major enabler.
“It’s using data to deliver the right care, at the right time, with the right dose.”
– Anna Stevens.
Plenary Three gave speakers a peek into how such a system works in the real world. They already exist and are being refined for both state and federal government health departments and private insurers across Australia.
HoneySuckle Health is a specialist healthcare data science and services company, a joint venture between healthcare provider Cigna Corporation and private health insurer nib.
CEO of Honeysuckle Health Rhod McKensey discussed how data analytics could be used to create better health outcomes. The impact of health management programs that insurer nib delivers to members can be better understood using data science, he told the seminar.
Using data science, Honeysuckle found that nib’s health management programs could be made more cost-effective through selective member targeting, which is critical to properly allocating scarce health resources.
Improving the cost-effectiveness of providing healthcare is a central focus of governments throughout Australia.
Ernst and Young presented a simulation model to support the provision of long-term healthcare planning for the Northern Territory Health Department, the Federal Department of Health, and New South Wales.
The model uses using a modular structure of demand, supply, and sustainability. The demand module uses data to analyse emerging population health dynamics, and the supply module examines how to meet those health needs with existing services. The third module of sustainability looks at the costs and benefits of delivering those health needs in the most cost-efficient and sustainable way.
Yet many speakers noted that while the amount of healthcare data was exploding, there is currently limited capacity to analyse it, and in some cases, the systems to analyse it are still embryonic.
Plenary Four provided some startling figures to the audience, demonstrating the amounts of health data that are currently being generated.
Stuart Houston of Google Health revealed that nearly 2 trillion medical images are created globally every year.
To apply machine learning over that data to analyse it requires a massive expansion of both storage capacity and analysis, he said.
The data itself is unorganised, siloed within different areas of one organisation, and reliant on multiple differing standards. There are challenges connecting data between different systems within an organisation, as well as across different organisations.
While machine learning is needed to analyse this data for the benefit of healthcare providers, it is proving challenging to get that data into the right shape, so machine learning can be applied to it, Henry Wei of Google Health told the audience.
Machines can be taught to learn and analyse, but questions remained around what those machines are learning. There are observer biases to consider, as well as issues of what questions were excluded, or whether patient answers weren’t recorded.
The algorithms that govern machine learning are still being trained to understand what is going on in the real world, and that is creating significant challenges for the healthcare industry.
In Plenary One, Dr Dana Kai Bradford, Principal Research Scientist at the CSIRO, provided some confronting examples of human-digital interaction gone wrong.
After warning the audience that they may find some of these examples distressing, she revealed that a French company conducting research into suicide discourse between humans and chatbots found chatbots struggled with deciphering idioms and vague language to identify suicidal ideation, Dr Bradford says.
The healthcare industry, including insurers, must think through the role devices will play in the future as they become more intelligent, as they could become liable in future legal cases, she says.
These were only some of the opportunities and barriers to providing better mental health treatment in Australia. The session showed how much further technology has to advance to provide more positive digital interactions with customers suffering from mental health issues.
The mental health problem is only set to grow.
Nearly 5.9 million Australians are at risk of a mental health issue, and the mental health impact of the bushfires of last year, COVID-19, and repeated lockdowns are not even fully understood yet.
While data analysis plays a central role in reducing healthcare costs for many economies in the OECD, Plenary Two stepped outside the Australian experience to examine what other approaches may be used in making healthcare more efficient. Notably, it examined the thinking behind the Risk Equalisation System in the Netherlands, as a prompt to encourage broad thinking in the current review of the Australian system.
Before introducing the Netherland’s Health Care Act in 2006, a typical insurer focused on risk selection, pricing, and underwriting. The Act outlawed underwriting, as is the case in Australia. The government had also intended that insurers would still have an incentive to improve efficiency through managing healthcare costs and service quality. The system equalises risks using a prospective method – i.e. equalising health risks across insurers based on next year’s expected costs rather than last year’s actual costs. The calculations use a wide range of data, assembled centrally by the government – but not available to insurers – and models designed and executed by the National Institute of Healthcare.
CPD: Actuaries Institute Members can claim two CPD points for every hour of reading articles on Actuaries Digital.