Wearables, Covid-19 and the Health-Tech Revolution

Covid-19 is an extraordinary but not unforeseeable crisis. Infection surveillance, infodemiology and state-based early-warning systems came up short, and for most governments containment via lockdown has been the only policy option available. But ultimately it is a blunt, economically damaging tool that creates a range of unintended consequences, the brunt of which are often borne by marginalised groups. For many countries, the immediate crisis is beginning to subside and a combination of measures, including testing and tracing, social distancing and the use of masks, are being applied as part of efforts to ease lockdown. 

This paper explores the role that wearable technology could play in improving early-warning systems for detecting Covid-19 infection and looks ahead to applications for these devices in a wider health-tech revolution.

Significant gaps remain in our understanding of Covid-19, and in many ways the coming period is going to be one of trial and error for nations. In particular, the asymptomatic nature of the virus in its early phase means that there is always a risk of unknown transmission. Governments will need effective trigger mechanisms – using novel data sources – and agile and responsive systems that enable them to act quickly if the reproduction number creeps up. Individuals may also want to take on greater responsibility for their health and disease detection, given the near universal failure of countries to price in risk and prepare for a pandemic, despite numerous warnings. With the power to both aggregate data and empower individuals, wearables could form part of a solution. 

1. Government should fund clinical studies to accelerate understanding of the utility of wearables as an early-warning system for Covid-19 and other viral infections. It should also support efforts underway, working to promote the utilisation of data for infection alerts. 

2. As part of a pilot phase, frontline workers should be provided with wearables that can detect heart rate and other biomarkers that can indicate illness. Vulnerable and marginalised groups should also be targeted if initial pilot phase is successful, in particular those that have higher risk factors as a result of age or comorbidities

3. Ultimately, government guidance might also consider public adoption of devices that have received regulatory approval and can be effective for early monitoring of infection beyond just Covid-19. 
    

The accelerating nature of technology over the past decade has meant that, compared to other recent outbreaks such as MERS and SARS, states have more tools at their disposable than ever before. Countries such as South Korea and Taiwan, who had been through the experience of infectious disease in the recent past, had adapted elements of the state to help with the response, building on their nations’ openness to technology to provide public services. But this has been slower in many legacy systems – including in the UK, which despite deep skills and expertise in this area, has faced institutional barriers to reform. However, the pace of change in recent weeks has been extraordinary in relative terms. Bold initiatives that would have taken years under normal circumstances – including the NHS Platform, which provides real-time actionable data for clinicians; a complete shift to telemedicine in GPs; and the provision of Facebook Portals for remote monitoring in care homes – have all been implemented. Part of the credit has to go to Health Secretary Matt Hancock and the NHSX, which was founded last year. But at a time when there is common collective will and endeavour happening across a range of issues, government should be going further and, where there is absorptive capacity within the system, be more open to reform.

As we have set out in previous papers, three key principles should guide technology policy in this regard:

1. Turn the networked public into an advantage.

2. Lean in to innovation and experimentation.

3. Be more transparent with the public than ever.

They also provide a framework for thinking about wearables.

The potential for wearables in health care has long been apparent. The Apple Watch, which primarily started out as a luxury complementary product to the iPhone, is today marketed as a health device. Similarly, Google, last year acquired FitBit for a deal that valued the company at $2.1 billion. Other companies are also innovating in this space, such as Oura (which tracks activity via a ring) and WHOOP (which uses a wristband). Apollo Neuroscience has developed a wearable to improve heart rate variability (HRV); based on research at the University of Pittsburgh, it sends gentle waves to increase HRV and parasympathetic activity, which can assist the body when it is under stress. This has been a growing market, with around 16 per cent of the US population alone owning a smartwatch (in the UK, around a quarter of the population own a wearable). The potential benefits of wearables are augmented by their increasing ability, the accuracy of the data they can collect, and the capabilities in Machine Learning and Artificial Intelligence to provide insights from that data.

At the most basic level, common consumer wearables measure fitness statistics such as heart rate. But technology has advanced significantly in recent years, and home wearables can now track metrics such as body temperature, blood oxygen levels and respiratory rate. Wearable devices can even aid diabetes patients by taking blood glucose readings and can monitor fertility. For example, those suffering from type-1 diabetes have formed an Open Artificial Pancreas System movement. Unable to get data from her continuous glucose monitor (CGM), founder Dana Lewis used code to obtain it and send notifications to her phone. Using this data, she developed an algorithm to predict the impact of food, insulin and activity on blood glucose levels over time. She then released an open-source version so that others could automate an insulin pump to keep blood glucose in a target range.

Devices that can take around 250,000 measurements per day have an even broader utility; over the course of a couple of weeks, they can collect enough information to calculate a baseline in physical metrics such as heart rate and temperature, then alert the user when there is a deviation from it. For example, average temperature of 37°C or 98.6 °F (range: 36.2–37.5°C [97.2- 99.5 °F]) was identified by the German physician Carl Wunderlich in 1852, based on observations of 25,000 patients. However, a recent study has found that the average temperature has dropped in the US since the Industrial Revolution, as have those in the UK. Building a greater understanding of ranges and averages at a population level are therefore important, but more critical is the need to understand individual levels. In doing so, individuals and physicians can be alerted when there are marked differences from personal norms, meaning decisions can be made off your individual metrics, rather than the current model of population-based measurements.

Beyond temperature, the range of metrics that wearables can collect includes:

• Heart rate

• Breathing rate

• Abnormalities in sleep patterns

• Blood glucose levels

• Muscle bio-signals (EMG)

• Blood oxygen levels/saturation

Blood oxygen level, for instance, has received significant attention during the Covid-19 pandemic. A number of patients who have contracted pneumonia as a result of the virus have also reported drops in blood oxygen level (a healthy range is 95 to 100 per cent) without experiencing any difficulty breathing. This has resulted in a large number of oximeters being purchased for home readings, although there has been some caution about the utility of such devices without clinician guidance. However, if there is good remote monitoring infrastructure in place, such data should be a useful tool.

The Apple Watch’s hardware has been able to take such readings since 2015, but the degree of accuracy is unclear and it has not received Food and Drug Administration (FDA) approval as yet. It is alleged that it may be part of the next release, which could provide consumers with more information that can then help guide clinical advice. (There have also been some examples of smartphone apps attempting to provide such readings, but the scientific basis has been deemed questionable.)

This highlights an important point of distinction between different health metrics: As it stands, many simpler measurements can be determined by consumer devices such as the Apple Watch, whereas others require more sophisticated technology that may only be available in a hospital. Even so, the analytical possibilities for some of the simpler measurements means they can provide extremely valuable insights.

Figure 1

Figure 1 – What can wearables potentially measure?

As with all infectious diseases, early detection is key to monitoring the spread of Covid-19. In turn, this will save lives. The difficulty with Covid-19 is that some cases can be mild or asymptomatic – either in the initial phase or throughout the course of the infection – so without effective monitoring systems, it can be incredibly hard to control transmission. The result has been the blunt option of lockdown.

It is possible, though, that data collected from wearable devices can improve our pandemic response by detecting early signs of Covid-19 cases. By monitoring variations in heart rate or body temperature, wearables could help to alert people to the possibility of infection. As a result, the data these devices collect could provide people with information on which to base decisions to isolate or to seek medical advice or treatment. Data from wearables can also enhance remote patient monitoring, which takes pressure off health-care systems and prevents unnecessary exposure of medical professionals to the virus, thus reducing transmission. Aggregate data taken from wearables can also contribute to research by detecting general patterns and trends within a population, which can contribute to improved public-health responses. Cumulative data can also be used to identify geographical Covid-19 hotspots.

There is still a need for clinical validation and an evidence-base in this area, and a number of researchers have launched studies. At the Stanford Healthcare Innovation Lab, Professor Michael Snyder (chair of the Department of Genetics at Stanford Medicine) is leading research that uses data collected from a variety of wearables, including FitBit and Apple Watch, and a series of algorithms that will be trained to detect whether an individual’s immune system is acting up. The work is based on research by Snyder and a former postdoctoral student, Xiao Li, which detected infection from changes in heart rates, recorded on a smartwatch. Data from the study is expected soon, which the team hopes that the algorithms will be able to detect Covid-19, even when carriers are asymptomatic. There is also potential that its use can be extended to infection alerts and even to diagnosis and prognosis. Scale and diversity of participants is however key for accuracy and representativeness. From a public policy perspective, if the algorithm is proved to be effective, it will require regulatory approval before it is deployed more widely.

A similar effort is underway in Germany, where public health authorities have launched a smartwatch app in partnership with a health-care startup Thryve. The aim of the app is to monitor the spread of Covid-19 and analyse whether measures to contain the virus are effective. The app gathers measurements on pulse, temperature and sleep to analyse whether users have Covid-19 symptoms; the government calls for citizens to donate data, much as they would blood, providing digital specimens to assist innovation and discovery.

Within the hospital setting, there have been further applications for wearables. For example, at the John Radcliffe Hospital, Oxford, researchers have been testing and developing the concept of a virtual High-Dependency Unit (vHDU). Here, high-risk patients are monitored using wearable sensors that measure pulse rate, respiratory rate and blood oxygen saturation, combined with Bluetooth-linked tablet computers and smart alerting algorithms. The aim of the exercise is to both reduce the burden on nursing staff and to improve early detection of abnormal physiological parameters. The technology has been modified and is now being applied to the isolation ward for Covid-19 patients who do not need to be ventilated. Imperial College and NHS organisations are also trialling a wearable sensor, developed by Sensium, to monitor people in quarantine and alert medical professionals if their health is deteriorating.

Together, these examples show the potential of wearables in health care – not least as these are already products in use by consumers. However, there are a number of challenges that need to be overcome, some of which are technical, some social, and others political. To summarise these:

Technical

• The need for clinical evidence: Empirical research on the clinical utility of wearables is still limited, as it is a relatively new technology. Many health-care providers are therefore waiting for further validation studies before implementing wearables in clinical settings. However, with greater commitment by governments and individuals to collect and donate data, the utility increases.

• Interpreting data: Depending on the feedthrough mechanisms into clinical environments, there is a risk of overwhelming clinical providers with large amounts of data from multiple sources. False positives are also likely, but effective processing and monitoring systems can help to alleviate these, in particular if the shift to telemedicine forms the basis for deeper reforms to health-care infrastructure. It can reduce the friction and cost of time with a medical practitioner, but also create a more integrated and remote, real-time care.

• Privacy/security: As with all data transmitted over digital networks, there is a risk of security breaches, which are particularly sensitive with regards to patient information. Important questions to address include where data from devices is stored, who has access, how long it is kept for and others. However, the collection of such data will be at a user’s discretion, so responsibility for how it is used would be user-defined.   

Social

• Access and availability: Internet and device access are not universal, although the data collected from a representative cohort can have positive sum effects on the wider population. Given the relatively low cost of some devices, there is also a role for state provision to frontline workers and vulnerable groups.

• Digital literacy: Operating wearable devices requires a degree of digital literacy, although alert functions can notify providers or friends and family. Wearables could be particularly useful in monitoring the health of elderly populations, yet generally this group is less experienced with technology.

• Interpretation of results: There is a risk of alerts creating anxiety, but usage is optional and results do not indicate diagnosis. For many people, understanding more about one’s own personal health and risk of infection will be beneficial, and the broader social effect can be positive.

Political

• Regulation: A number of barriers prevent the wearable-technology industry from undergoing further innovation, with devices often subject to complex and lengthy regulation. In the US, however, during the Covid-19 crisis, the FDA has issued a new policy that allows manufacturers of certain FDA-cleared devices to expand their use so that health-care providers can use them to monitor patients remotely.
  
  But there is also precedent in this space: The Apple Watch’s electrocardiogram (ECG) function has received clearance from the FDA as well as from 19 European regulators, including those in France, Germany, the UK and Italy. In Europe, new Medical Devices Regulations have also been delayed as a result of the crisis, while in the UK, the Medicines and Healthcare products Regulatory Agency has been fast-tracking some regulations where appropriate.

Wearables can provide a key early-warning system about the likelihood of Covid-19 infection, but their use can potentially go further in infection surveillance. For example, a study published in The Lancet in January this year by Scripps Research used FitBits with sleep- and heart rate–tracking capabilities to predict influenza-like illness better than the Centres for Disease Control and Prevention (CDC). Using de-identified data from the devices, such as resting heart rate and sleep data, the researchers could improve real-time predictions at the state level; by contrast, the CDC relied on data from outpatient health-care providers, which could mean a delay of one to three weeks.

An inability to “nowcast” has hampered responses throughout the Covid-19 crisis, but real-time predictions should be key to early-warning systems in the future. They can mean pharmaceutical and non-pharmaceutical interventions, such as physical distancing, can be rapidly applied. For the flu alone this could be significant: In the UK, 18,000 deaths were attributable to this virus in 2017 and 26,000 the following year. Worldwide, it accounts for about 650,000 deaths per year. The wider effects of flu outbreaks, both in terms of wellbeing and knock-on effects to the economy, are also significant.

From an individual perspective, wearables can alert you to changes in your biometric data so you can take action, whether that be through rest or through contacting medical professionals, who can provide guidance on the best course of action. As Stanford’s Michael Snyder has said, “An alert isn’t a direct diagnosis, and it will be important for folks to be able to contextualize their situation and use some common sense.” Through the right consent mechanisms, de-anonymised data can also be aggregated into platforms, so that policymakers, clinicians and epidemiologists can analyse it and use it to inform their decisions. As we’ve seen through Covid-19, crowdsourced data-aggregation efforts are also springing up and could form a strong independent basis for trusted monitoring.

To further preserve privacy, decentralised mechanisms could also be used; Machine Learning models such as Federated Learning allow data to be analysed without leaving an individual’s device. No individual data is therefore collected on a server. Other technical breakthroughs, not least in encryption, also should shift the contours of the privacy debate. But together, these also highlight a further failing in health care that pre-dated the crisis: Our inability to use technology to turn the scale of biomedicine into an advantage.

We have failed to find ways to open up data to increase our understanding of disease pathogenesis, classification, diagnosis, prevention, treatment and prognosis. But a confluence of advancements – including a deeper understanding of biology on a systems level through the omics, gene-editing, and our ability to interpret data through Machine Learning and Artificial Intelligence – can change the shape of health-care delivery. It should be personalised and focused on the individual, so they have greater control and understanding, but it should also aggregate so that we treat health as a common collective goal. Wearables have the potential to be one part of this system, and policymakers should be exploring their use with far greater interest today. As the price of these devices continues to fall and the consumer demand increases, so does their potential utility.

If health systems had accelerated the adoption of technology available over the past decade, the magnitude of the current pandemic would likely have been much less severe. In response to Covid-19, reform has accelerated across many vectors of the health system already, with institutional barriers being broken down and innovation increasing. But the momentum needs to increase. Society cannot be stuck in a cycle of lockdown while the world waits for a vaccine. 

New and innovative ideas need to be explored, with wearables having the potential to fulfil a vital function in early-warning systems. By providing continuous, real-time physiological information via dynamic, non-invasive measurements of biochemical markers, they can help monitor issues – from identifying signs of potential infection (as with Covid-19) to assisting with broader care of diseases such as diabetes. They can also be part of a wider positive long-term shift in health care, so that it is more personalised and more precise, and so that individuals have far greater control and understanding of their own physiology and what changes to it mean.