Loading field map
Digital health is the field of practice and study concerned with the use of digital, mobile, and wireless technologies to improve health and healthcare. It encompasses the design, development, evaluation, and implementation of tools that collect, store, analyze, or transmit health-related data, as well as the systems and policies that govern their use. As a subfield of health informatics, digital health is distinguished by its focus on technologies that are used directly by patients, consumers, and clinicians in everyday settings, rather than on the internal information systems of healthcare organizations alone. Its central questions revolve around how these technologies can make care more accessible, personalized, and efficient, and how to ensure they do so safely, equitably, and without adding burden to clinicians or patients.
Digital health is a broad umbrella term that includes mobile health (mHealth) applications, wearable sensors, remote patient monitoring, telehealth and telemedicine, personal health records, patient portals, clinical decision support tools delivered at the point of care, and the data platforms that integrate information from these sources. It also includes the use of artificial intelligence and machine learning to interpret the large volumes of data these technologies generate.
The field is driven by several enduring questions. One is whether digital tools genuinely improve health outcomes, rather than merely increasing the volume of health data or the convenience of accessing it. Another is how to design technologies that are usable and beneficial for diverse populations, including older adults, people with low digital literacy, and those with limited access to smartphones or broadband. A third question concerns the integration of patient-generated data into clinical workflows: how can clinicians trust, interpret, and act on data collected outside the clinic without being overwhelmed by it? Finally, digital health grapples with regulatory and ethical questions about data privacy, security, and the potential for algorithmic bias to worsen health inequities.
The roots of digital health lie in earlier efforts to use computers in medicine. From the 1960s onward, hospitals and researchers developed clinical information systems, electronic medical records, and decision-support tools. These were largely institution-centered, requiring dedicated hardware and trained operators. The term "telemedicine" emerged in the 1970s to describe the use of telecommunications for remote diagnosis and consultation, often over dedicated video links between hospitals.
The modern digital health landscape took shape in the late 2000s and 2010s, driven by three converging developments: the widespread adoption of smartphones, the miniaturization of sensors, and the expansion of high-speed wireless networks. These made it possible to collect health data continuously and unobtrusively in daily life, and to deliver interventions directly to individuals. The rise of consumer wellness devices—such as activity trackers and smartwatches—created a large market for personal health monitoring, while the growth of app stores enabled the rapid distribution of health applications. Around the same time, the U.S. and other countries introduced financial incentives for healthcare providers to adopt electronic health records, which created the data infrastructure onto which digital health tools could be attached.
The COVID-19 pandemic in 2020 marked a significant acceleration. Lockdowns and infection-control measures forced rapid adoption of telehealth for routine consultations, and many regulatory barriers to remote care were temporarily relaxed. This period demonstrated both the feasibility and the limitations of digital care at scale, and it permanently shifted expectations about the role of remote and asynchronous care.
Digital health is not organized around a single paradigm but rather around several distinct approaches that address different problems and operate with different assumptions. These approaches coexist and often overlap, and their boundaries are porous.
One major tradition treats digital health tools as medical interventions that must meet the same standards of evidence as drugs or devices. This approach, rooted in evidence-based medicine, emphasizes randomized controlled trials, clinical endpoints, and regulatory approval. Its practitioners—often academic researchers, clinical epidemiologists, and regulatory scientists—ask whether a given app or wearable actually improves blood pressure control, reduces hospital readmissions, or changes health behaviors in measurable ways.
The strength of this approach is its rigor. It protects patients from ineffective or harmful products and provides clinicians with trustworthy guidance about what to recommend. Its limitation is that the pace of clinical research is slow relative to the speed of technology development. By the time a trial is completed, the product being tested may have been updated or discontinued. Moreover, the controlled conditions of a trial may not reflect how people actually use these tools in their daily lives, where engagement is often intermittent and context-dependent.
A second tradition comes from human-computer interaction, user experience design, and participatory design. Its central concern is usability and engagement: a digital health tool that people do not use, or cannot use correctly, cannot help them. This approach emphasizes iterative design, user testing, and co-design with patients and clinicians. It asks not only whether a tool works in principle, but whether it fits into the routines, capabilities, and preferences of its intended users.
This tradition has contributed important insights about the "last mile" problem in digital health—the gap between a technically functional tool and one that people actually adopt and sustain. It has also highlighted the risk that poorly designed tools can widen health inequities, if they assume high literacy, fluent English, or comfortable access to technology. Its limitation is that good design does not guarantee clinical effectiveness; a beautifully designed app may still be based on a flawed theory of behavior change or may not address the underlying medical condition.
A third approach centers on the computational analysis of health data. This tradition, which draws on machine learning, statistics, and signal processing, treats digital health as a data-rich environment in which patterns can be discovered that would be invisible to traditional clinical reasoning. Its practitioners develop algorithms to detect atrial fibrillation from wearable sensor data, predict deterioration from continuous monitoring, or identify individuals at risk of diabetes from electronic health records combined with consumer data.
This approach has expanded the possibilities of what can be measured and predicted, and it has driven the development of new sensor technologies. Its central challenge is the gap between predictive accuracy and clinical utility. An algorithm that predicts a heart attack with high statistical accuracy may not be actionable if clinicians do not know what to do with the prediction, or if it generates too many false alarms. There is also the risk of algorithmic bias: models trained on data from one population may perform poorly on another, and the opacity of some machine learning methods makes such failures difficult to detect.
A fourth tradition focuses on the integration of digital tools into complex healthcare systems. This approach, rooted in implementation science and health services research, recognizes that a digital health tool is not a standalone intervention but a change to a sociotechnical system. Its practitioners study how telehealth programs are adopted by clinics, how patient portals change the division of labor between clinicians and administrative staff, how remote monitoring affects the workload of nurses, and how reimbursement policies shape the sustainability of digital services.
This tradition emphasizes that the success of a digital health tool depends on factors that have little to do with the technology itself: workflow redesign, training, leadership support, financial incentives, and alignment with existing clinical culture. Its limitation is that implementation research is often context-specific, and findings from one setting may not transfer directly to another. It also tends to be descriptive rather than prescriptive, offering explanations for why implementations succeed or fail rather than simple formulas for success.
These four approaches are not rivals in the sense of competing for exclusive truth. They address different parts of the same problem. The clinical validation approach establishes whether a tool works; the human-centered design approach ensures that people can and will use it; the data science approach extracts meaning from the data it generates; and the implementation approach determines whether it can be sustained in real-world settings. A mature digital health project typically requires all four.
However, there are genuine tensions. The slow, cautious pace of clinical validation can frustrate designers and data scientists who work in rapid iteration cycles. The data science approach's appetite for large, continuous data streams can conflict with the human-centered tradition's emphasis on user control and privacy. The implementation tradition's attention to local context can seem to undermine the generalizability that clinical researchers seek. These tensions are productive: they force the field to confront the fact that digital health tools are simultaneously medical interventions, consumer products, computational systems, and organizational changes.
The current landscape of digital health is characterized by several durable features. First, the field is highly commercialized. Many digital health products are developed by startups and technology companies rather than by academic medical centers, and their business models often depend on direct-to-consumer sales or on partnerships with employers and insurers. This creates a dynamic in which marketing can outpace evidence, and in which products may be designed to maximize engagement rather than clinical benefit.
Second, regulation is uneven. In many jurisdictions, low-risk wellness products are largely unregulated, while products that make medical claims or that are intended for diagnosis or treatment are subject to oversight. The boundary between wellness and medical use is often blurred—a smartwatch that detects irregular heart rhythms may be marketed as a consumer product but functions as a screening device. Regulators have developed expedited pathways for software as a medical device, but the pace of regulatory adaptation lags behind the pace of innovation.
Third, the field is marked by persistent inequities in access and benefit. Digital health tools require devices, connectivity, and literacy that are not evenly distributed across populations. There is growing evidence that these tools can amplify existing health disparities if they are adopted primarily by younger, wealthier, and more educated users. Addressing this requires attention not only to the design of the tools themselves but to the social and economic conditions that shape their uptake.
Fourth, the field has moved toward integration and interoperability. Early digital health was characterized by a proliferation of standalone apps and devices that did not communicate with each other or with electronic health records. The current direction is toward platforms that aggregate data from multiple sources, application programming interfaces that allow different tools to share data, and standards that enable interoperability. This integration is necessary for digital health to move beyond isolated experiments and become a coherent part of healthcare delivery.
Finally, the field is increasingly attentive to the problem of evidence generation itself. Because traditional clinical trials are often too slow and expensive for digital products, researchers are developing alternative methods: pragmatic trials embedded in routine care, single-case designs, and the use of real-world data from the tools themselves. These methods are promising but raise their own questions about bias, confounding, and the reliability of data collected outside controlled conditions.
Digital health remains a young and rapidly evolving field. Its core insight—that the widespread availability of computing power and connectivity can transform how health is monitored, maintained, and restored—is now well established. Its ongoing challenge is to realize that potential without sacrificing rigor, equity, or the human relationships that remain central to care.