Why watch wearables?
16 Sep 2015
There is a huge buzz around health wearables. Andreas Caduff, Biovotion, discusses the challenges – from skin connection to data collection – of getting them right.
A recent author in this Swiss Re series, Karin Frick of the Gottfried Duttweiler Institute, stated: “A health revolution is coming. This much is clear. Through smart devices and smart analytics we will soon have personalised health predictions. What is less clear is what we might do with this information.” Just for that reason alone – we all need to watch the wearables.
A brief search on Google’s NGram Viewer suggests the hype for revolutions (red for health revolution, blue for healthcare revolution) may be over; but wearables are going from strength to strength (green). The Apple Watch, for example, has captured the imagination of many journalists. Analysts are projecting large increases in sales for the medical wearable market from humble starting points today.
It seems prudent to try to understand what is driving such enthusiastic projections, and assess whether these drivers are sustainable.
What, if anything, justifies this attention?
A wearable may confer value as an accessory or enhance the status of its wearer. Even if for the latter purpose, the wearable still has to measure something. Measuring alone does not add value if that measurement does not lead to, at least in principle, a resulting action. Taking action can affect health in at least three different ways:
Physiological tracking and subsequent corrective action can lengthen lifespan and quality of life. For instance, the protocols for managing cystic fibrosis have led to dramatic extensions of the lifespan, following ever more refined protocols in a disciplined fashion. This is an extension of classic healthcare provision into the home. Even in a pre-disease state, monitoring the physiological state may allow early detection of the onset of a condition which can either be averted (such as pre-diabetes) or managed better. This may be relevant for a consumer wellness market, or an extension of a clinical market, depending on local conditions.
Even if vital signs monitoring does not double lifespan in all individuals or prevent a metabolic syndrome reliably, providing the measurement can empower the individual to take care of their own health. It gives a strong signal of caring about physical well-being. Vital signs monitoring may significantly improve customer loyalty even where health gains are less obviously tangible.
These value endpoints existed 20 years ago and have not really changed. While there has been ongoing refinement of how to effect change in physiological states, this has also not been a game changer. Rather, there are three main factors which together suggest why the market for medical wearables might have arrived with force:
Market demand has grown and is growing. With an increasing number of elderly people and an increase in the non-communicable disease burden, chronic conditions, which require constant monitoring of vital signs, are becoming more common.
There is a growing supply of medical wearable devices which meet the standards of consumer lifestyle devices. The second is the creation of consumer lifestyle devices, monitoring a suite of variables that might be described as 'wellness' related. The rapid spread of lifestyle devices has seen cross fertilisation with medical devices. One element has been design, something the traditional medical device industry was only passingly interested in. Another has been simplicity. Medical devices used to require calibration and set-up. Increasingly they follow the click and play model. A further adaptation has come in functionality. Whereas medical device manufacturers were previously able to instruct their users (typically the professional care provider) on exactly how to use their device, they must increasingly follow consumer devices, where devices may be used in ways not originally foreseen or expected by the manufacturer.
Rapid technological progress now allows integration of medical wearables into different healthcare settings. For instance, wireless transmission and big data interfaces allow constant connectivity and open the way for health data to be analysed by health care providers as well as by users and consumers. This allows for wearable vital sign monitoring to move between different settings of care seamlessly and effectively, enabling ‘continuous care’ not just a continuum of care.
Figure 1: Cycle of care
What to watch in a specific medical wearable
Despite what many may think, wearable devices are not new. They have been around for many years in the medical space, used to monitor a number of vital functions. Any wearable medical device needs to fulfil five criteria:
Use case(s) with real demand
Sensors and algorithms to reliably measure physiological change
Attachment to the body of one device, preferably with multiple sensors embedded
Actionable interface based on the measured change
Integration with the health ecosystem
1. Use case(s) with real demand
For a specific medical wearable to take off requires both a significant need for monitoring physiological change to manage the condition, and a matching willingness to pay for it – irrespective of who is the end-payer.
The most attractive markets are targeted at health conditions which affect a large fraction of the population. These require frequent or continuous use, as in the case of many chronic conditions, not least because it generally requires buying rather than renting the wearable device. Devices are designed to work together with medical professionals rather than replace them. The new ability to charge for remote monitoring under Medicare in rural areas provides a significant incentive to clinicians to use medical wearables in home monitoring. For providers who receive fixed reimbursements for treating a condition, clinicians need to be enabled to be providing better quality care with higher efficiency. For instance, if doctors know which of their patients need more urgent attention, they can prioritise their time better. Such event-driven prioritisation makes care delivery more responsive and efficient.
2. Sensors and algorithms to reliably measure the physiological change
A key challenge is to be able to measure with clinical grade accuracy. A good signal to noise ratio is critical if one is to rely on the information for decision-making. This is a bar that is challenging for many existing wearables. Sensors must work for all sort of people, irrespective of size, gender, skin colour – and should work out of the box.
Sensors and associated algorithms should enable measurement regardless of what daily life throws at them. A vital sign monitor for an athlete is unlikely to be useful during exercise if it cannot handle motion or sweat. This robustness should, of course, be underpinned with the necessary clinical studies.
Most sensors cannot be uni-purpose. Several vital signs need to be measured simultaneously. Consumers do not want multiple sensors hanging onto their arms. Measuring with multiple sensors need to be related to each other and the data fused.
The next phase in the successful deployment of wearable devices will be to improve their predictive capabilities. Ultimately algorithms will utilise sensor signals to anticipate acute conditions; the onset of a chronic condition; or the longer term deterioration of that condition. First generation technology will be able to suggest conditions such as Alzheimer's, asthma, depression or sleep disorders. Second generation technology will monitor cardiac pressure, weight, blood pressure and fluid status for suggestions of heart failure. Third generation technology will be able to measure blood glucose and haemoglobin. All these functions will be integrated into one single device.
3. Attachment of the sensors to the body
Attachment to the body has wrongfully received relatively little attention. A wearable needs to be worn, and if measurement is to be continuous over time, it needs to be comfortable to wear for extended periods. This often conflicts with the need to have a good skin contact, however, which is vital if sensors are to generate hospital grade clinical data. Devices on the wrist would have to be excessively tight for most users if vital signs are to be captured with an acceptable degree of reliability. In addition, during use with patients with chronic conditions, it is important for materials to be hypoallergenic and well-tolerated. This issue has caused product recalls in the past. A more comfortable, discrete and robustly measurable location has been identified in the upper arm. Figure 2 shows a multi-sensor device which can currently capture six bio-signals: heart rate; blood oxygenation; skin temperature; blood pressure wave; steps/motion; and respiratory rate. In the more distant future, sensor devices may well be based on technology that can be ingested, or monitors an individual’s condition at a distance, without needing to be worn. For instance, there are apps that can derive heart rate from a facial video, by analysing colour changes as a result of blood pulsation.
Figure 2: Biovotion monitor
4. Actionable interface
Raw data from a device only becomes useful if processed in a system that allows for meaningful resulting actions. What is actionable information varies significantly between use cases. Sometimes, simply seeing the vital sign is sufficient. For instance, one of the factors that differentiates people who successfully lose weight and keep it off, is that they check their weight daily.
If measurements are continuous, and many parameters are measured, viewers are likely to be overwhelmed by data. Where this is the case, it may help to first transform a multiplicity of signals into a richer, single dimension. An extreme version of this would be a translation into a red-amber-green traffic light. There are also more information-rich transformations, such as the display of a probability density of multi-dimensional states (see for instance this as demonstrated in this video by Biovotion).
Regardless of the transformation applied to the data, it helps if it is appropriately contextualised. This can be as simple as showing the data compared to benchmarks with others or self over time, or against some health target. Good user interface design clearly makes a big difference, even once the information to be shared has been optimised.
A primary care physician who is home-monitoring a panel of chronic condition patients is not likely to find a continuous stream of vital signs data particularly actionable – even if wonderfully transformed and contextualised. More likely is that predictive algorithms will prioritise which patients are likely to require attention. Within the hospital, this is a classic situation faced in intensive care units, where there is continuous vital signs monitoring. This frequently results in false positives, leading physicians to sub-optimally allocate their scarce time.
Prioritising and alerting the user through predictive algorithms is not just an issue for clinicians in clinical settings, but has significant potential in enabling individuals and patients in their own homes to take better, healthy action.
An actionable interface could conceivably go even further, if there is a very strong relationship between certain vital sign values and actions to be taken. In late stage cases of chronic pulmonary obstructive disease (COPD), for instance, patients require extra oxygen, but not too much. A closed-loop system could show where the external oxygen supply is driven mechanistically by the vital signs measures, such as oxygen saturation and patient activity.
5. Integration with health ecosystem
The processing of the data relies on the existence of effective information eco systems. First data from the wearable needs to be able to enter the ecosystem. Data must be transferred seamlessly and wirelessly from the wearable sensor to a monitor (typically a Bluetooth enabled phone or handheld device); from the monitor to a cloud (for secure data evaluation and sophisticated functionalities); and from cloud to a viewing platform (a computer or tablet).
If the data is simply for personal use, it is relatively straightforward with today’s technology to create a solution. Handling privacy and security is a major requirement even in this restricted, data silo approach. To begin to unlock the promise of continuing care, the wearable data must be made available in a host of different formal care settings securely, automatically and in a trusted fashion.
Dependable wearable sensors will be of growing importance in hospitals. Patients in intensive care are constantly monitored as a matter of course. Once out of intensive care, such monitoring can be continued with such devices, combined with sophisticated codified expert knowledge and support in returning patients home more quickly – all while still being monitored. For this to occur, devices need to be constantly connected; need to operate on the same platform as hospital systems; and capture all the critical data that could be recorded in a hospital. One study of 18,000 patients in 2012 measured an increase in survival rate of 6.3% with a wearable device; a decreased length of average hospital stay by 3%; and a saving of 1.7 nurse hours per day . For major hospitals with thousands of staff, these represent very large savings. Another study utilising predictive health monitoring with codified expert knowledge combined with vital sign monitoring even succeeded in completely eliminating avoidable deaths .
The litmus test for health care systems of the future will be to bring together the various sensor reporting devices, together with point of care diagnostics. This must be done in a way that can interact and contribute to a patient's electronic record. It can be combined with sophisticated software based automation. Sensor devices will not be fulfilling their potential if their data stays isolated in silos.
1. A controlled trial of electronic automated advisory vital signs monitoring in general hospital wards.
Bellomo R, Ackerman M, Bailey M, Beale R, Clancy G, Danesh V, Hvarfner A, Jimenez E, Konrad D, Lecardo M, Pattee KS, Ritchie J, Sherman K, Tangkau P; Vital Signs to Identify, Target, and Assess Level of Care Study (VITAL Care Study) Investigators.
Crit Care Med. 2012 Aug;40(8):2349-61. doi: 10.1097/CCM.0b013e318255d9a0.
2. Cardiorespiratory instability before and after implementing an integrated monitoring system.
Hravnak M, Devita MA, Clontz A, Edwards L, Valenta C, Pinsky MR.
Crit Care Med. 2011 Jan;39(1):65-72. doi: 10.1097/CCM.0b013e3181fb7b1c.
Andreas Caduff was a speaker at the Centre's conference Transforming healthcare: Telemedicine, best practice and you.
Download full article (PDF 53 KB)