At the end of November, AliveCor’s Kardia Band became the first FDA cleared medical device accessory for the Apple Watch – converging lifestyle tracking with medical advice. The device uses artificial intelligence to assess whether the user’s heart rate is in the predicted range based on their activity – captured by sensors in the Apple Watch (i.e. light sensors to determine heart rate and accelerometer data used to calculate activity). If it’s out of sync, then the user will be advised to take an electrocardiogram (ECG) reading by placing their thumb on a silver patch located on the wrist band.
The device detects abnormal ECG traces, information from which can be used by a medical professional to diagnose atrial fibrillation (AF). Since patients with AF have an increased risk of stroke and other serious complications, earlier diagnosis and treatment may help to prevent or control the morbidity associated with AF. The seamless integration of the device into daily life through a wearable (watch) enables data to be collected continuously, possibly preventing the devastating consequences from undiagnosed AF.
The big data picture
The Kardia Band is not the only device or data platform that combines disparate data streams to make actionable predictions. There is growing recognition that the amalgamation of different health data types could reveal the most valuable insights. Already some hospitals in the UK (e.g. Salford Royal) are exploring the inclusion of self-tracking data into electronic healthcare records with the ambition of informing new care models. Whilst current efforts are focused on integrating data captured by health-tech (including mobile apps and wearables) that the individual intentionally interacts with, there is a growing interest in utilising other sources of citizen generated data (CGD) for health-related insight. This includes the vast amount of data individuals now produce about themselves through internet services, social media, and mobile phones. Although not originally created to deliver health related analysis, the potential of this data to inform health research is beginning to gain traction.
For example internet searches are being investigated for tracking communicable diseases (e.g. Dengue) and the wealth of information on social media is being explored for predicting mental health decline (e.g. predictive markers for depression by analysing photos posted by users on Instagram). Additionally, researchers are exploring the use of speech records to diagnose disease (e.g. Parkinson’ diseases) and predict likelihood of suffering severe mental health decline (e.g. predicting psychosis episodes in high-risk youths). These advances could be harnessed through voice-controlled tools, such as ‘intelligent assistants’ like Siri, Google Assistant and Cortana, which can capture and store users’ voice clips. The enormous challenge, however, lies in constructively integrating these new insights into health practice for use by individuals to manage their health or by healthcare professionals to inform patient care. But these exciting developments beg the question – will citizens want their data to be used in these ways?
The enormous challenge, however, lies in constructively integrating these new insights into health practice for use by individuals to manage their health or by healthcare professionals to inform patient care
Are we leaving the ‘unquantified’ behind?
There is also the question of the digitally ‘left behind’ those people who do not interact with such technologies and therefore, have a relatively small digital emission. This could be due to a range of factors including – differences in interest in health technologies, personal preferences, income to access consumer-facing health technologies, or variance in technical aptitude to make use of internet and digital tech.
According to Ofcom’s yearly report, use of the internet, social media and smartphones among over 65s, is increasing. However, they remain one of the least digitally engaged demographics. Furthermore, findings suggest that even if they do interact with these technologies, many over 65s do not feel confident, particularly when managing personal data. This raises the question of if and how this demographic will interact with the increasing range of wearables and personal tracking devices with growing number of features and sensors.
Minding the data gap
Beyond the variable access to, or use of these digital tools that could potentially help improve health, the unequal representation of certain demographics within citizen generated datasets may also have knock-on effects for others in that population. For example, data analysis and interpretation tools such as machine learning (ML) based analytical approaches which are developed using existing data, may not work effectively for those populations not sufficiently represented in the ‘training’ datasets. This could potentially mean that future health tools may not serve those that are currently under- or unquantified. So, despite the excitement surrounding the use of CGD for health, it is clear that there are number of considerations – from data gaps, to user interest and expectations- that will need to be addressed along the way.
A new PHG Foundation project is looking into the challenges presented using citizen-generated data, more information on which will be available on the website soon.