Loading field map
In the 1960s, a patient in a rural clinic with a puzzling rash or an abnormal heart tracing had few options beyond a telephone call to a distant specialist. The specialist could hear a description but could not see the rash or read the tracing. That practical barrier—health care trapped in the same physical room as the patient—launched the first digital health framework. Over the next six decades, the field expanded far beyond remote consultation, adding patient-controlled information, mobile interventions, software that acts as medicine, continuous sensor streams, and algorithms that learn from the resulting flood of data. Each new framework did not simply retire the old ones; it layered on top of them, creating a stack of tools that now coexist, compete, and depend on one another.
Telemedicine began as a solution to the access problem. Early projects in the 1960s and 1970s used closed-circuit television and dedicated phone lines to connect a specialist at a teaching hospital with a primary-care clinician in a remote location. The core commitment was synchronous, clinician-to-clinician consultation at a distance. The framework was narrow by design: it preserved the traditional doctor-patient hierarchy and did not give patients direct control over the technology. Over time, the label broadened into "telehealth," which includes direct-to-patient video visits, remote patient monitoring, and store-and-forward imaging. Yet the original model—a clinician using a communication channel to deliver care across space—remains the backbone of the framework. Telemedicine did not disappear when later frameworks arrived; it became infrastructure. Today, a wearable sensor or a mobile app often feeds data into a telehealth platform, and the COVID-19 pandemic dramatically accelerated its adoption, making video visits routine in many health systems.
The rise of the internet in the 1990s created a new pressure: patients who could search for health information online wanted more than a brochure from the doctor's office. Consumer health informatics responded by building tools that put health information and decision support directly into the hands of patients. Online health portals, personal health records, and interactive decision aids for treatment choices all belong to this framework. Its distinctive move was to shift agency from the clinician alone to the patient as an active participant. Unlike telemedicine, which kept the clinician as the gatekeeper of the interaction, consumer health informatics assumed that patients could understand and act on health information if it was presented clearly. The framework coexists with telemedicine—a patient might use a portal to message a clinician before a video visit—but it also narrowed over time. As mobile phones became ubiquitous, many of the functions that consumer health informatics pioneered (symptom checkers, medication reminders, health education) were absorbed into mobile health apps, leaving the framework's core as the design and evaluation of patient-facing information systems.
Mobile health, or mHealth, emerged alongside consumer health informatics in the 1990s but took a different technical path. Where consumer informatics was built for desktop web browsers and later adapted to phones, mHealth committed from the start to mobile devices—first SMS-based interventions, then smartphone apps. The framework's central contribution was action-oriented, real-time intervention. A diabetes management app could prompt a patient to log a blood glucose reading immediately after a meal, not just when they next opened a computer. mHealth narrowed the scope of consumer health informatics by focusing on behavior change in daily life rather than general health literacy. It also transformed telemedicine's remote monitoring: instead of a clinician calling a patient to ask for a blood pressure reading, the patient's phone could collect and transmit the data automatically. The framework's methods include ecological momentary assessment, push notifications, and gamification. Today, mHealth is the most widely deployed digital health framework, but its evidence base is uneven. Thousands of apps exist, yet only a fraction have been tested in controlled trials, creating a tension that later frameworks would try to address.
By the early 2000s, the proliferation of health apps raised a question: could software itself be a treatment, not just a tool to support treatment? Digital therapeutics answered yes, but with a crucial condition—the software must meet the same evidentiary standards as a drug or medical device. This framework commits to delivering evidence-based therapeutic interventions through software, typically requiring clearance from regulators such as the U.S. Food and Drug Administration (FDA). Unlike most mHealth apps, which may claim to help with stress or sleep without rigorous testing, a digital therapeutic must demonstrate efficacy in randomized controlled trials (RCTs) and receive regulatory authorization for a specific indication. Examples include prescription digital therapeutics for substance use disorder, insomnia, and attention-deficit/hyperactivity disorder. Digital therapeutics narrows and formalizes the behavioral intervention tradition of mHealth: it takes the same idea—software changing behavior—but demands a higher evidence threshold. It also overlaps with the discipline-root framework of Clinical Decision Support, because both involve software that makes treatment recommendations, but digital therapeutics is unique in being the treatment itself rather than a tool that advises a clinician. The framework remains in active tension with mHealth over what counts as sufficient evidence, and with payers over reimbursement models.
While mHealth and digital therapeutics relied on user-initiated actions—opening an app, tapping a button—wearable and sensor-based digital health shifted data collection to continuous, passive monitoring. Starting in the 2000s with consumer fitness trackers and later expanding to medical-grade patches, smartwatches, and implantable sensors, this framework generates streams of physiological data (heart rate, activity, sleep, glucose, oxygen saturation) without requiring the user to do anything. The key difference from mHealth is the mode of data capture: mHealth asks the user to log or report; wearables sense automatically. This shift created new opportunities and new problems. On the positive side, wearables can detect arrhythmias, falls, or early signs of infection that a person might not notice. On the negative side, the data are noisy, prone to artifacts, and raise privacy concerns that earlier frameworks did not face. Wearables also blurred the boundary with telemedicine: a clinician can now monitor a patient's heart rhythm remotely through a wearable sensor, turning telemedicine's episodic consultation into continuous surveillance. The framework's data streams became essential infrastructure for the next framework, data-driven digital health, because machine learning models need large, high-frequency datasets to train on.
The most recent framework, data-driven digital health, emerged around 2010 as electronic health records, wearables, mHealth apps, and telehealth platforms accumulated vast amounts of clinical and behavioral data. The framework's commitment is to use machine learning and big-data analytics to discover patterns, predict outcomes, and personalize interventions at a scale that rule-based systems cannot match. Where earlier frameworks designed fixed protocols (a digital therapeutic delivers the same sequence of modules to every patient), data-driven digital health adapts dynamically: an algorithm might predict which patients are at risk of hospital readmission and trigger a preventive telehealth check, or it might analyze wearable data to adjust a diabetes management plan in real time. This framework depends on the data infrastructure built by all its predecessors—telemedicine provides the clinical context, mHealth and wearables generate the behavioral and physiological streams, and digital therapeutics supplies the evidence standards for validation. The relationship is one of layering and absorption: data-driven digital health does not replace the earlier frameworks but integrates and extends them. Its challenges include algorithmic bias (models trained on one population may fail in another), regulatory uncertainty (how to approve a model that changes as it learns), and the need for interoperability standards that allow data from different devices and health systems to be combined.
All six frameworks remain active today, and their coexistence creates both synergy and friction. They agree on several broad principles: digital tools can improve access to care, enable continuous monitoring, and personalize treatment in ways that traditional clinic-based models cannot. They also agree that evidence generation matters—though they disagree sharply on what counts as good evidence. Telemedicine and digital therapeutics tend to demand randomized controlled trials, while mHealth and wearable advocates argue for real-world evidence, pragmatic trials, and continuous evaluation as data accumulate. Regulation is another fault line. Digital therapeutics has embraced FDA clearance as a gold standard; mHealth and consumer informatics have largely operated outside regulatory oversight, though that is changing as the FDA's Digital Health Innovation Action Plan extends its reach. Data ownership divides the frameworks as well. Wearable and mHealth companies often claim ownership of user-generated data, while consumer health informatics traditionally assumed that patients control their own health information. Finally, the question of agency—who acts on the data—remains unresolved. Telemedicine and digital therapeutics place the clinician in charge; consumer health informatics and mHealth empower the patient; data-driven digital health can tip either way, depending on whether the algorithm's output goes to the patient's app or the clinician's dashboard. These disagreements are not signs of failure. They reflect the fact that digital health is not a single technology but a layered ecosystem of frameworks, each optimized for a different part of the care delivery problem, and each still evolving in response to the others.