Despite the promise of smartphones and wearables for early detection of cancer, there are several hurdles to implementation that require attention. Key barriers to success are outlined here, alongside suggestions for how these may be addressed with future research.

The use and adoption of smartphones and wearables in this context require robust research into the selection of a signal, the development of sensors, and the generating evidence of accuracy and utility of those sensors in the real world. This includes identifying physiologic (or emerging pathophysiologic) signals that are most predictive of cancer, determining how often these need to be collected, and elucidating what other data would add precision to results (eg, age, risk factors, and presence of symptoms). For technology developers, sensors are usually designed and prioritized for a number of potential applications, mainly targeting overall health and wellness rather than diagnostic capabilities per se. Prioritizing these research and development efforts for cancer detection specifically over and above other priorities could be challenging to justify for business development reasons. Relatedly, the original intended commercial purpose of existing sensors may not have been connected to cancer detection. To make headway in this field of research, technology developers and device users will need to be willing to provide access to data for research. General Data Protection Regulation allows device users to share their data with third-party organizations under the right to portability. Developing systems to facilitate data sharing, in formats compatible with health data, could allow the generation of new data sets to signal cancer risk. This will prevent duplication of effort and maximize the use of existing data for public benefit.

While initial evidence on the accuracy of a sensor to detect a given health signal could involve case-control studies (eg, individuals recently diagnosed with cancer and matched controls), subsequent research would likely require large prospective cohorts. Further, given the weak signal-to-noise ratio, it is likely that signals from wearables or smartphones alone might lack sensitivity or specificity. Therefore, research that combines signals from wearables or smartphones with other digital sources of data (eg, symptoms recorded in health records and initial laboratory tests in primary care) will almost certainly be needed to demonstrate sufficient accuracy and utility in target populations. This was recently highlighted in a systematic review [29] of artificial intelligence technologies for skin cancer detection. Despite an abundance of digital products, the review highlighted that there has been very little testing in low-prevalence populations and limited data on the use of lower-quality images (eg, taken by patients or family physicians or using lower-quality phones), and as such, widespread adoption into practice has been limited [30].

Innovators also need to consider (and test) whether these new digital tools should and could detect more than 1 type of cancer (or detect other potentially important nonmalignant diseases; eg, cirrhosis in individuals with jaundice or depression in individuals with weight loss). Other considerations include who the target group is (eg, all adults or only those at higher risk of developing cancer), at what point in time (eg, certain age), and at what periodicity that group should begin using this technology for the detection of cancer. Further, it is well documented that symptom monitoring increases selective attention to the body, resulting in increased symptom reporting [31]; thus, the monitoring of symptoms could result in increased symptom burden. Development and testing need to determine the extent to which the monitoring of activity, symptoms, and other signals changes the outputs of those measurements [32,33].

Given the burden involved in fulfilling regulatory approvals for diagnostic devices, many smartphone technologies and wearables that could potentially have value for cancer detection will instead be introduced as products for overall health and wellness management. As the field expands, more guidance and standards for digital health tools are being introduced to ensure that they are not only safe and effective but also adoptable by the health care system [34]. For example, the recent UK National Institute of Health and Care Excellence (NICE) evidence standards framework [35] is intended to ensure new digital health technologies are clinically effective and offer value to the health care system. The framework includes standards concerning safety, quality, acceptability, bias mitigation, data practices, professional oversight, credibility with health professionals, safeguarding assurances, scalability, as well as evidence of real-world performance and use. In some countries, consumer protection regulations also determine standards that certain wellness features (eg, step counting and heart rate measurement) need to fulfill, even though these are not regulated medical devices. As specified in the CanTest framework for early cancer detection [36], research and development will benefit from this early specification of the criteria (eg, target product profiles) needed for successful digital products for cancer detection [37-39].

A key issue of wearables and smartphone technologies is the potential for new innovations to exacerbate inequalities in cancer outcomes. Sociodemographic factors such as household income, age, level of education, and gender have been found to influence the use of mobile health (mHealth) technology [40-44] and there is “a real risk that the increased use of digital technologies will make care experiences and outcomes worse for some people (or communities)” [45]. Development of wearables and smartphone technologies for cancer detection should be conducted with an equity lens to focus on the views and needs of those living or working in more deprived areas and those at risk of lower health literacy (eg, those with lower educational level, older age, lower income, and ethnic minority groups) [46,47], so that cultural attitudes toward the use of technology, affordability, and access can be a focus in their development. Inclusion and diversity within the development and testing of sensors are vital so that products are not biased and work equally regardless of skin color or other physiological differences [29]. Affordability is also a crucial point. Even though smartphone use is extensive [48] in both higher- and lower-income countries, the availability and quality of sensors differ across brands and models of smartphones. Wearable devices have far lower penetration in most high-income countries and lower still in those individuals with lower socioeconomic status. If accurate and reliable sensors are only available on high-end devices, then the net result will be inequitable outcomes. The consideration of a reverse innovation approach may be useful here if it is possible, to focus testing on inexpensive, easy-to-use products that can be rolled out at scale.

Crucial to the successful implementation of any innovation is early insight into the user perspective, including acceptability, uptake, and correct use by the intended target group [49-51]. Yet, this consultation is often omitted or occurs too late in mHealth implementation, resulting in user burden; technical issues; poor designs; and ultimately the lack of uptake, adherence, and impact of the technology [49,52-54]. Indeed, there is currently an absence of research on user perspectives on wearables and smartphone technologies for the detection of cancer. While key issues such as cost, motivation, comfort, ease of use, trust in data use, visibility, and interpretability of data are applicable across the spectrum of wearables and smartphone technologies in health [40,55], there may be additional, specific challenges for using these innovations for the detection of cancer.

In this context, acceptability also pertains to individuals’ willingness to share their data from smartphones and wearables with researchers, medical professionals, or private companies. Willingness to share data from wearables was reported to be lower than that for other commercial data [56]. Less than 15% (n=65) of survey respondents aged 60 years and older were willing to share wearable device data with academic research institutions and only 40% (n=423) of those aged 18-59 years were willing to do so. Trust in organizations and worry about data misuse have been shown to be a key factor in people’s willingness to share commercial data for health research [56]. Ensuring clear and transparent data use and data-sharing policies is vital for success. There are real concerns about the misuse of data, commercialization, and access to data by unauthorized people [57]. While data sharing is an essential component in the use of smartphones and wearables for cancer detection, data protection is equally as vital.

In research studies of wearables, dropout rates can be up to 44% [40], and nonadherence to wearing devices for the study duration can be up to 50% [58,59]. Nonadherence is likely to be even higher in people with preexisting comorbidities and for technologies requiring long-term engagement, as may be needed for cancer detection to track signals over time. Balancing the advantages and disadvantages of continuous versus intermittent measurements at certain intervals should be a key consideration.

It is, therefore, essential to investigate user perspectives in parallel with the development and potential future deployment of wearables and smartphone technologies for cancer detection. This also includes encompassing the views of clinicians who are involved in the ongoing surveillance and care of those with a history of cancer and would inevitably be involved in shared decision-making on the potential implementation of such technologies and, crucially, the ongoing clinical management of individuals whose sensors indicate signals of possible cancer. In general, primary care clinicians have not typically been deeply engaged in the implementation of other consumer-grade or regulated medical devices; understanding from these clinicians’ viewpoint on how they could use information from smartphones and wearables within their clinical care pathways is critical to any adoption [60].

The exciting potential of wearable technologies for cancer detection must be considered alongside the possible negative consequences. As seen with other new developments in cancer detection, given the overall very low prevalence of cancer, even tests with very high specificity will lead to a large number of individuals with false positives. The subsequent need for investigation and resources needed to differentiate those with false versus true positives (ie, do have cancer) could be considerable. For the majority of individuals, this could lead to huge risks of overinvestigation and inappropriate use of health care resources [61]. For cancer detection specifically, we can anticipate a far higher potential for wearables and smartphone technologies to generate anxiety than for other conditions (eg, detection of sleep apnea, or detection of irregular heartbeat), especially among those already fearful of cancer recurrence. This is particularly relevant to the question of how “results” should be delivered to users, what support would be needed at that time, and whose responsibility this would be. On the other hand, wearable use may lead to a false sense of reassurance, leading to a lower perceived need to attend cancer screening or respond to symptoms (eg, “my wearable says I am healthy...there is no need to see my doctor”). This is similar to when a negative cancer screening test result can overly reassure patients and affect subsequent decisions to seek care [62]. These issues about the psychological and behavioral impact of smartphone technologies and wearables to detect cancer remain unexplored and need focused behavioral science research.