Can smart devices do something more than just show us time?

A short overview of the accuracy of smart devices for health applications

An athletic looking person holding and watching smart devices: a smart watch and a smartphone

Many of us have been using a smartwatch or a fitness tracker device for some time now, either as an extension of our smartphone devices or for health and athletic related reasons. Some others may be using a smart scale, and other smart versions of conventional devices, such as smart devices to measure the blood glucose levels, body temperature and blood pressure.

Consequently, a question that people often ask is whether those devices can really do something more than just show us nice visualizations of some measurements. Is it possible to detect symptoms of a yet undetected infection? Could they be used to predict potential health issues? Can they identify cases where urgent medical care is required? Thankfully, many researchers have been working on answering those questions!

So, what kind of information can be collected by wearable devices? 

In [1] the authors present in detail different properties and how they could be measured. These include the now typical properties, for instance: heart rate, blood pressure, physical activity, but also others more advanced such as respiratory rate, sweat and even emotion, which can be strongly influenced by the sleep quality. The authors report that even if these devices are slow to enter the healthcare market, they present tremendous opportunities due to their capabilities.

Another feature mentioned by [2] is that these devices can provide continuous ECG monitoring, which can be uploaded online and assessed from anywhere in the world. This is a major advantage compared to conventional Holter monitors, which must be returned to the doctor for analysis. The authors also mention that even if there are some limitations regarding those devices, nevertheless, they remain optimistic by saying that these existing issues can be alleviated with the future developments and algorithms.

Current research

Even though these smart devices still may not be the ultimate tool for diagnosis of atrial fibrillation they can be used for the general population as a warning tool. These limitations mentioned by [2], were apparent at a large-scale Apple-funded research presented in [3], which tracked more than 400 000 people for 8 months. The goal of this study was to examine the possibility to detect atrial fibrillation using the commercially available Apple watch. The results, however, were not so consistent, they showed low sensitivity and specificity.

In contrast, an investigation trying to predict the risk of diabetes had better results, as shown in [4]. The researchers collected some basic information from the users such as height, weight and age, and from different devices they collected step count, type and duration of activity, calories, glucose levels, blood pressure and more. By analysing all these data and implementing different Machine Learning algorithms they had more than 80% accuracy in predicting diabetes risk!

An unusual case is reported in [5]. A seemingly healthy person received a notification from their smartwatch that their heart rate was irregular and relatively high. After being admitted to the hospital and undergoing some procedures, it was found that the irregular heart rate was caused by a previously undetected tumour in their lung, which was promptly removed. Without the warning from the watch, the researchers comment, that the tumour would have continued growing worsening his prognosis.

Advantages in a bigger scale

Other than that, smartwatches can even have advantages at a larger scale. In [6], researchers showed that by tracking various metrics from wearables, real-time influenza-like illness rates can be predicted, which can give the time to hospitals and medical institutions to adapt and prepare for increased influx of patients. Another development according to the authors is that this kind of analysis can be implemented in the future for the detection of influenza-like illnesses at an individual level. You could get a notification that you have the flu even before having any symptoms!

And what about COVID-19? Early results by [7], showed that using their method infection with COVID-19 could be detected 9.5 days before the symptom onset! However, they report that without additional measurements and data, COVID-19 could not be distinguished from other virus infections.

All these sounds very promising, are there no concerns or issues?

As mentioned before, the results from the large-scale Apple study [3] were not very consistent. Low accuracy and specificity lead to increased cases of false positive (healthy people receiving warnings for potential issues) or false negatives (people who have AF but is not detected by the watch).

In addition, people trust and can be misled about the capabilities of their devices. Therefore, without receiving any warning, false negative cases may avoid seeking medical health, increasing the risk on their health. On the other hand, false positive cases can lead to overdiagnosis, can induce unnecessary high medical costs, and lead to an overload to the healthcare system. 

Researchers commented on the reasons why the Apple study was not successful. The heart rate measurements were not so frequent, so short duration irregularities were missed. Additionally, these smart devices are designed to perform many different actions and therefore require frequent charging. Many people, in addtion, also avoid wearing their devices during sleep, a period which could potentially reveal a lot of useful information. That means that for hours at a time no measurements at all are made.

The authors of the study [8] are also a bit more hesitant about the use of these devices in medicine. According to them, the quality of the data is questionable, there can be unclassifiable recordings, which cannot be used and there are potential cyber security issues.

Future prospects

Nonetheless, one cannot help but be optimistic. Recent developments are trying to deal with the limitations of these devices. Newer devices can have continuous heart rate and ECG monitoring, and battery life of multiple days, making them capable of detecting short duration irregularities. Additional sensors such as blood pressure, skin temperature and oxygen saturation open up the possibility to accurately detect a variety of potential health issues. More people will be open to wear their devices at a more frequent basis and even during sleep, as these devices progressively become lighter and better fitting. Improved and more advanced algorithms will be better at analysing the collected data and extracting valuable information.

I personally cannot wait to see what the future brings!


[1] I. C. Jeong, D. Bychkov, and P. C. Searson, “Wearable devices for precision medicine and health state monitoring,” IEEE Transactions on Biomedical Engineering, vol. 66, no. 5, pp. 1242–1258, 2019, doi: 10.1109/TBME.2018.2871638.

[2] J. M. Raja et al., “Apple Watch, Wearables, and Heart Rhythm: where do we stand?,” Annals of Translational Medicine, vol. 7, no. 17, pp. 417–417, 2019, doi: 10.21037/atm.2019.06.79.

[3] M. v. Perez et al., “Large-Scale Assessment of a Smartwatch to Identify Atrial Fibrillation,” New England Journal of Medicine, vol. 381, no. 20, pp. 1909–1917, Nov. 2019, doi: 10.1056/nejmoa1901183.

[4] J. Ramesh, R. Aburukba, and A. Sagahyroon, “A remote healthcare monitoring framework for diabetes prediction using machine learning,” Healthcare Technology Letters, vol. 8, no. 3, pp. 45–57, Jun. 2021, doi: 10.1049/htl2.12010.

[5] A. P. C. Cole, A. Kar, K. Nimako, and J. Smelt, “Smartwatch helps detects lung cancer: Using personal technology to advance healthcare,” JRSM Open, vol. 13, no. 1, p. 205427042110686, Jan. 2022, doi: 10.1177/20542704211068651.

[6] J. M. Radin, N. E. Wineinger, E. J. Topol, and S. R. Steinhubl, “Harnessing wearable device data to improve state-level real-time surveillance of influenza-like illness in the USA: a population-based study,” The Lancet Digital Health, vol. 2, no. 2, pp. e85–e93, 2020, doi: 10.1016/S2589–7500(19)30222–5.

[7] T. Mishra et al., “Pre-symptomatic detection of COVID-19 from smartwatch data,” Nature Biomedical Engineering, vol. 4, no. 12, pp. 1208–1220, Dec. 2020, doi: 10.1038/s41551–020–00640–6.

[8] D. Jin, H. Adams, A. M. Cocco, W. G. Martin, and S. Palmer, “Smartphones and wearable technology: benefits and concerns in cardiology,” Medical Journal of Australia, vol. 212, no. 2, pp. 54–56.e1, 2020, doi: 10.5694/mja2.50446.

The article was written by Angelos Theofilatos. 

Contact Us

Patras Science Park S.A. Stadiou, Platani 265 04, Greece