This month’s story, Stealthcare by Liz Williams has two main themes, a form of residential care system provided on a cruise ship and wearable technology incorporated into a health insurance plan and lifelong care provision. The story focuses on an abnormality in the data stream from Rosalind Lee’s wearable health monitor, which has resulted in confusion about her insurance plan and the inclusion of an existing condition. Health surveillance and insurance systems have been a recurring theme through the Fictions series with Life’s Lottery, When I Visit the Palace, and George focusing on how real-time data and an AI model could underpin such a system. Where data analysis is explored for informing care provision, the stories often return to the idea of hacking or falsifying an individual’s data to game the system for better access to healthcare.
Wearable technology and health monitoring
In this series of short stories, we do not get a sense of how the various health systems in place were set up, or where the insurance model of healthcare comes from. Where data analysis and AI are being utilised, it is not clear if the approaches were built as part of a pre-existing insurance-based system, or if the design and implementation of such approaches results in the insurance model after deployment. As an example from another sector, the installation of “telematics” or black boxes for car insurance has changed how insurance premiums are calculated. This is an interesting approach to a domain where many drivers feel that premium models are unfair for young people and older people alike. In this instance black box telematics should make things fairer, although it may ultimately reveal little overall difference to previous population averages. However, on an individual level, such tailored approaches are likely to be fairer than current averages which affect entire groups equally. The NHS provides free healthcare at the point of use, and it is interesting to consider how monitoring or “telematic” devices could help provide higher quality, more equitably accessed care. Insurance based healthcare is a controversial subject in the UK, largely because of what we see of the US system of healthcare. It is unsurprising that many of the stories we commission have dystopian imaginings of a universal technical system underpinned by various forms of insurance. As technology and legislation develop, there are recurrent concerns raised about the security of the NHS and its data from different nations and their interests in developing products using said data. Our recent series of guest blogs on trade and discussion of the Trade Bill highlighted some of these concerns.
Many wearable devices designed for healthcare are designed to track health features and disease risks. Currently watches, smart rings or skin patches are the most advanced forms of health tracking. Over time many more forms of technology will emerge, we are already seeing prototypes of digital contact lenses and sensors embedded in tooth implants for health monitoring. Many wearable devices, which provide a steady stream of data, use a proxy measure of a biological feature. These can vary in accuracy and can introduce bias to data models and AI systems. A good example of this measurement inaccuracy introducing data biases is the measurement of oxygenation of people’s blood using a pulse oximeter. The pulse oximeter has been recommended by clinicians for home monitoring during COVID-19, with a blood oxygen saturation of 92% or lower requiring urgent referral to hospital.
However, Pulse oximeters are known to give false readings for individuals with darker skin, with the margin of 10% found in the study, this can be the difference between a referral to hospital for treatment and going untreated for a severe condition. Such issues need to be swiftly addressed as the medical and lifestyle wearable market expands, without doing so would exclude many people from access to better health outcomes. If we consider approaches beyond relatively simple light spectrum measures such as disease biomarkers or genetic data these biases already exist in the corpus of research and need to be corrected rapidly.
If we consider how such biases could relate to the story, the inaccurate readings from a wearable could be misunderstood to be a fraudulent attempt to hack or mask a trace. We discuss hacking of health data and other types of personal data used for surveillance in Getting Real: Personalised medicine, how personal is too personal?. In Stealthcare, alongside surveillance, we see that the insurance model has a form of “pay-to-play” modification of an insurance plan, where individuals can pay a higher premium to engage in behaviour which is riskier to their health. This model favours those who can afford it, even if the activity is overall worse for a person’s health. If we consider someone who may be genetically predisposed to a condition (for example heart disease), they could pay a higher add on premium (for example, a drinking or smoking add on) to someone who does not have such a risk factor. The premium cost would be a disincentive and the individual with a condition may subsequently live a healthier life. But the element of choice would be removed, and people may still choose harmful behaviour over a more cost-effective health plan. Public health policies are continually looking at ways to reduce unhealthy behaviours, to improve population health and reduce the financial burden on the NHS. In recent years smoking; drinking alcohol; consuming sugar, salt, and fat; and getting more exercise have been key campaigns. There is a developing debate about the harms of e-cigarettes, which have been a central part of the government strategy to reduce tobacco consumption. The success of such policies relies on population statistics and proxy data measures as well as survey data for assessment. Using wearables and other forms of data capture would be able to provide much more accurate outcome data. However, this surveillance based approach could dramatically reduce public trust.
The lifetime wearable we see in the story is relatively new as a healthcare concept and provides a pivotal detail which provides the cliff hanger at the end of the story. We have seen this type of wearable in previous stories and digital exclusion in a care setting was a focus of discussion in the associated Getting Real blog. It is interesting to consider the traceability of such wearables and an apparent switch that may have occurred at some point in Mrs. Lee’s life. Traceability of products and people is a complex technical challenge. We have seen in the last year some of the difficulties of keeping track of people moving in an environment through COVID-19 contact tracing applications. QR codes have made a bit of a comeback as a method to “check-in” as well as to provide a form of vaccine passport for international travel. When it comes to products or objects, tracing them and being able to identify them is useful to maximise use and monitor safety. If a product is faulty and needs recalling, it is important to be able to identify how many are deployed, where the products are and, if time sensitive, which products are a priority. Product identifiers are also vital to determine whether counterfeits are entering a market. This is particularly important for medicines and healthcare products. Barcodes, serial numbers, and embedded identifiers can be used for physical products. A key consideration in the developing digital healthcare product market is the identification of digital systems and keeping track of versions of software, as well as ensuring an update to an app does not inadvertently cause harm to patients. The problem has not yet been fully addressed by legislation, regulation, or standards. The main approach taken internationally is the use of unique device identifiers, with a new identifier being issued for software when key criteria are affected (including changes to; original performance, safety, interpretation of data, modified algorithms, database structures and architecture. See clauses 6.5.1-4).
Models of care provision
The story takes place in a cabin on a luxury cruise ship, which is where the main character lives, providing the function of a retirement property. This concept of using a cruise ship for retirement is one which has been considered several times as an alternative to care homes, with several different calculations of cost. In the story the main character has purchased a cabin, but in reality the cost of a round the world luxury cruise appears to be comparable to the cost of care in a residential home. For a cruise ship to be effective and cost efficient as a care environment it would likely need to have a different sort of on-board infrastructure. In the story, Mrs. Lee is approved for residence on the ship by her insurance company and we see that there are many facilities on board, which are a range of medical and well-being treatments.
The cost of care can be extremely high and the care system in the UK is incredibly complex, mapping out the pathways across the care sector and how they interact with clinical or primary care pathways is almost impossible. FCC trustees Andrew Whelan and Irene Gray recently wrote blogs to describe some of the complexities and challenges facing social care and the various choices families and individuals have to navigate at a time which is usually already challenging.
The innovation and integration government publication, leaked in draft in February, explored some of the ways health and social care can be integrated and improved through innovation. The draft White Paper set out some of the plans for developing an integrated care system (ICS) and has been criticized for a lack of detail in several areas, notably models of funding. A more joined up health and care service would no doubt improve the quality of care provided in health and social care settings. But there are concerns that new proposals and legislation introduced during a pandemic response will not be adequately scrutinized by clinicians and professionals delivering health and care services.