By Dan Jenkins & Paul Draper, DCA Design International
The devices we use on a daily basis are getting smarter.
Devices that were once ‘dumb’ are now fitted with a range of sensors allowing them to work out what is happening around them, and how they are being used. For example, the latest generation of toothbrushes are able to determine when, and for how long, we are brushing our teeth. What’s more, they can also evaluate our brushing technique, telling us if we are pressing too hard, or spending too much time in one region of our mouth.
Augmenting everyday devices with sensors, microprocessors, communication technologies, and algorithms provides the possibility to allocate tasks that were once the responsibility of the end-user to either a microprocessor and software programme or another human elsewhere. Automation can offer a clear benefit by taking on the tasks that humans often perform poorly at, or would prefer not to engage in. Activities like continuous activity monitoring, or providing timely reminders may be better allocated to a microprocessor.
That said, it is important to remember that automation also comes at a cost. In the majority of cases, the same functions still need to be completed; however, they are simply passed from a human to a microprocessor or another human. Where data collection and decision making is distributed, it is imperative to ensure that the necessary communication is possible and assess the frequency and costs of this.
For consumer devices, the latest technology is frequently used as a marketing driver. Smart products are often proposed as a greater value than a non-connected version. This often comes with the promise of the ability to control objects in our home remotely, or closely monitor what is happening and respond accordingly. From a commercial perspective, products are often made ‘smart’ in an attempt to encourage users into a wider ecosystem of products or services and collect rich and valuable data on user behaviour. In some cases, there is a strong push to upgrade products to smart devices without explicitly considering the additional benefits for the consumer.
While the value of these connected smart products is not always immediately clear to all of us, the demand is often unquestionable. Well-considered and well-designed connected devices have the potential to optimise the allocation of function and improve the overall system’s performance in terms of efficacy, efficiency, safety, inclusiveness, satisfaction and flexibility.
For medical devices, adding intelligence to the device is often cited as a significant opportunity for enabling more patients to take control of their therapies. The idea of reminding users when to take their drugs, and recording what was taken and when is clearly appealing. Likewise, the ability to collect rich, often continuous, data on biometrics (e.g. heart rate, blood glucose levels, blood pressure) can be invaluable. This has the potential to allow patients the ability to manage conditions in the home that once had to be handled by HCPs. Similarly, it provides HCPs with a more robust evidence-base with which to make a diagnosis and track patient conditions.
Given these inherent advantages, there is a clear appetite to develop connected medical devices, albeit with some reservations about their implementation.
The role of smartphones
Smartphones have become the middleman in our relationship with the majority of the smart devices we interact with. Most smart consumer products take advantage of the advanced processing power, storage capacity, and relatively large, high-resolution, screens of our smartphones. By connecting these smart consumer devices to the ‘super-computers’ in our pockets, the connected devices themselves can be kept relatively simple and cost effective.
For consumer products, shifting the ‘intelligence’ to the app (running on the smart phone) has a number of clear advantages.
1. It reduces the bill of materials cost of the smart device in terms of processing power, memory, and display.
2. It allows systems to be updated easily (via updates to the smart phone app) without having to make changes to the device.
3. User interfaces and even functionality can be highly customisable.
4. The smart device app can exchange information with other apps on the phone to gain greater contextual understanding (e.g. location, weather, calendar, health apps).
For medical devices, however, the picture is somewhat different, as different constraints are placed on the system. The two most obvious differences are:
1. The requirement for regulatory approval.
2. The development time of a medical device.
The impact for smart medical devices is that one has to question if the model used for consumer goods, where the intelligence resides in the app, remains fit for purpose. The advantages of a lower bill of material cost remain appealing; however, new challenges are introduced.
Firstly, the challenge of proving that the software is safe and effective is far more demanding if the software sits within a complex operating system. For any app, running on a mobile phone or similar, controlling a medical device it would need to be demonstrated that the function of the app and data integrity cannot be corrupted by the operating system. Phone and tablet operating systems tend to have major updates every 12 months with numerous smaller updates through the year. Each of these updates may require the app software to be updated. Furthermore, each time the app software is updated the system will need to be assessed for new risks and may need further regulatory approval.
With medical device development timelines covering multiple years, it’s unlikely that the operating system or even the smart phone, that the medical device is designed to work with at the start of the project will be the same as the one at launch. It is far more likely that there will be multiple changes of phone operating systems throughout the development process. What’s more, given the long development time and the high investment, medical devices are often expected to remain in the market for longer than consumer devices.
Thus, the intuitive answer is that medical devices need to be smart, much smarter than their consumer counterparts. Allowing them to be far more independent of the phone they may be connected to. Any interaction with smart phones and tablets needs to be carefully considered. Ideally, any integration with a phone would be non-critical to the function of the medical device, reducing the regulatory approval burden for the app and the phone.
A smart approach to smart device development
Just like all connected devices, the first stage of developing a smart medical device should involve a detailed consideration of its purpose and the potential value of the connected system, above the legacy non-connected system. An explicit consideration should be made for the end-user and each of the stakeholders in the system.
An understanding of the information requirements should underpin the design of the system. This involves determining what information is required, when it should be displayed, where in the system, to whom and how (in what format).
In summary, the architecture of a connected medical device should be informed and driven by a combination of stakeholder needs, technological capability, appropriate risk, and the ability to gain and maintain regulatory approval. If a decision is made to allocate some tasks or functions away from the user to a ‘smart’ device, in the majority of cases, it makes sense that the intelligence lies in the physical device rather than the app. Apps can often offer a secondary view of this information; however, the regulatory overhead is likely to be reduced if the processing takes place on the device.