Source: The Lancet
The primary way of identifying symptoms of impending disease, and when it is time to seek care, relies heavily on a patient's own perception of how they feel. This reliance on a disease reaching a threshold at which subjectively recognised symptoms are produced creates a situation in which many patients identified as at risk for disease are already experiencing chronic disease. Thus, substantial delays can be seen from the first noticing of symptoms to accurate diagnosis. The COVID-19 pandemic has highlighted the inadequacy of subjective approaches for early detection of disease and control of the spread of infectious diseases. The capabilities of wearable devices and smartphone sensors to collect high-dimensional objective information and return this information to patients offers an opportunity to disrupt our current ways of detecting, classifying, and treating disease. Yet, current definitions of, and approaches used to detect, signs and symptoms of disease are misaligned with these new digital approaches.
Symptoms encompass the subjective sensation of the presence of disease: the patient must be aware of the symptom. Objective measures of health are considered signs observable by doctors, such as a skin rash, blood pressure reading, or biomarker of disease. Thus, rooted in this language, symptoms reported by patients must be subjective in nature whereas an objective measure cannot be a symptom. An important nuanced layer in these definitions is the role of the doctor—the expert who must be the observer and translator of objective signs of disease. Definitions aside, if patients were able to collect accurate objective measures of health themselves, and could track these measures independently, would these not be patient-generated signs and symptoms that could be adequate grounds for action?
New technology that allows patients to monitor measures of health that traditionally were defined as objective signs of disease, such as blood pressure, resting heart rate, heart rate variability, and activity, are changing the symptom reporting landscape. Up to now, the usual metrics tied to digital devices included physiological measures of activity (eg, steps or movement). These metrics are not clinical symptoms making them not so useful in their raw form. Electrical signals from digital sensors augmented by machine-learning techniques could be translated into momentary, minute-level, day-to-day, week-to-week fluctuating streams of high-dimensional health information. Multimodal approaches that translate passive device sensor datastreams to symptoms, which have been historically measured and understood through a subjective lens (eg, mood and cognition) are being developed.
Further, rapid development of apps for this purpose is accelerating progress, such as the Share the Journey app
for breast cancer, and the mPower app for Parkinson's disease.
The goal of these approaches is to translate complex sensor data into clinically meaningful patient-generated symptoms and signs that might improve individual early detection and surveillance and aid more efficient clinical decision-making.
These new capabilities to detect objective measures of health that can be returned to individuals raise important questions on what constitutes a symptom. Symptoms could be detected without a patient's awareness whereas objective signs could be detected without a doctor's involvement or expert translation. Our old definitions of subjective symptoms and objective signs of disease simply do not work in the digital era.
New digital approaches also disrupt the traditional power balance between patients and doctors. Artificial intelligence enables devices to learn complex patterns in day-to-day shifts in behaviour, cognition, emotion, physiology, and—with advancements in wearable biosensors—other measures of neuroendocrine activity.
The ability to return this health information back to patients incites questions around the shifting role of the patient versus the medical expert. Indeed, discussion on the shift to participatory medical models calls into question our current top-down biomedical approaches, realising the powerful insights that can emerge from individuals themselves. This new digital environment accelerates participatory medical models, enabling patients to work alongside clinicians and, occasionally, take the lead in the management of their own health.
Current medical models assume that patients can only parochially consider subjective aspects of their disease and not the objective facts needed by the doctor to make proper medical decisions. The prerequisite loop that requires the doctor to provide facts about a patient's disease impedes progress towards participant co-driven health and is misaligned with the new capabilities of the digital era. Digital technology enables patients to feed their own generated objective health information into decision trees, making it reasonable to assume that patients should be able to trigger medical decision making outside the current prerequisite loop.
Finally, new technologies enable potential detection of early disease manifestations not typically associated with subjectively detectable disease activity. This possibility has powerful implications for the area of infection diseases with respect to objective early detection of infection. For example, studies utilising wearable devices are showing promise in early detection of COVID-19 in asymptomatic individuals who show changes in heart rate, activity, and sleep.
How will new digital ways of detecting and tracking disease fit into health systems and the everyday life of users? This integration will undoubtedly comprise a concerted and challenging effort that could revolutionise the health of individuals and populations.
Approaches are still in an early phase, with much validation work being needed. Extensive pilot testing of the feasibility, acceptability, validity, and reliability in the use of these tools in patients, health-care providers and most importantly groups that are less likely to engage to avoid widening health inequalities is needed alongside rigorous data governance models. Potentially, the largest obstacle to progression of digital health approaches lies in compatibility with existing medical approaches, classifications of disease, and mistrust in readouts from digital devices. Ensuring the readout from digital devices is accurate and interoperable will be a first step in removing this obstacle. To make individualised and patient co-driven care a reality, a major shift is needed with respect to how patients are enabled to report on their health and, in parallel, how we conceptualise evidence of disease.
All authors declare no competing interests related to the perspective presented here. JRG is supported by the National Institute of Health Research (NIHR) Oxford Health Biomedical Research Centre (grant BRC-1215-20005). The views expressed are those of the authors and not necessarily those of the UK National Health Service, NIHR, or the UK Department of Health.