As the technology behind wearables for healthcare evolves in its sophistication, there’s a lot of interest in applications for remote monitoring, including for clinical trials. Wearables offer the potential to reduce the cost of these trials and drug development. There’s plenty of skepticism about the execution, particularly when it comes to issues like ensuring the accuracy of the data. That hasn’t deterred healthcare companies from studying these devices.
An article from Applied Clinical Trials fleshed out some of the challenges associated with using devices, such as sleep monitors, accelerometers and other tracking tools, and in some cases offers some suggestions for how to deal with them. Here are a few of them.
The context of data
This is a pretty broad issue that comes up repeatedly, but the article highlights the problem of sleep monitors, as an example. They could be particularly useful for tracking asthmatics because when they wake up in the middle of the night it is frequently due to an attack. Or it could just mean they need the bathroom. Or had a bad dream. There’s no way for a sleep monitor to contextualize it because it notes each one as the same. Although patients may remember a few, they probably won’t recall every single attack. To get more context, in-person clinical assessments are useful, such as the 6-minute walk test for people with COPD. The distance that a patient can walk in 6 minutes — either using a treadmill or an empty corridor circuit — is recorded, to get a controlled assessment that would be used for context on what the patients can achieve. Although the author recognizes the benefit of observing in person how patients do using their activity trackers, there are still going to be several factors that contribute to good and poor performance so it’s impossible to get complete context. Ultimately, the author concludes that the benefits of having more data to evaluate from tracking patients remotely “might mitigate against additional random error associated with this approach.”
When should inactivity be written off as non-wear time? It’s an important issue and will be different for each patient population. Defining a period that’s appropriate for the patients being evaluated is critical. As the author points out, “Selecting the appropriate threshold will ensure accurate identification of periods of non-wear and differentiate these from periods when the patient was not active while wearing the device.”
As the article points out, the amount of energy it takes one patient to do any one task varies dramatically depending on each patient’s condition. Translating counts into Metabolic Equivalent of Task or kilocalorie expenditure may not be accurate in patients with, say, COPD, or other illnesses that undermine fitness. Calibration equations tend to be based on relating activity to energy use in healthy users. So someone with COPD may be using much more energy to accomplish certain tasks than a healthy person. Is it possible to get more contextualized data for these patients? That’s likely to be one area scientists are looking at.
There’s a certain amount of randomness that needs to be factored in to tracking people with wearables that can account for lost data. If patients just aren’t feeling well enough to do exercise, they may not see the need for wearing an activity tracker. Or they’re doing an activity for which they can’t bring wearables along, such as swimming