Review

Wearable technology and the cardiovascular system:

the future of patient assessment
Gareth J Williams, Abdulaziz Al-Baraikan, Frank E Rademakers, Fabio Ciravegna, Frans N van de Vosse, Allan Lawrie, Alexander Rothman,
Euan A Ashley, Martin R Wilkins, Patricia V Lawford, Stig W Omholt, Ulrik Wisløff , D Rodney Hose, Timothy J A Chico, Julian P Gunn, Paul D Morris

The past decade has seen a dramatic rise in consumer technologies able to monitor a variety of cardiovascular Lancet Digit Health 2023;
parameters. Such devices initially recorded markers of exercise, but now include physiological and health-care 5: e467–76

focused measurements. The public are keen to adopt these devices in the belief that they are useful to identify and Department of Infection,
Immunity and Cardiovascular
monitor cardiovascular disease. Clinicians are therefore often presented with health app data accompanied by a
Disease, (G J Wiliams BMBS,
diverse range of concerns and queries. Herein, we assess whether these devices are accurate, their outputs validated, A Al-Baraikan MSc,
and whether they are suitable for professionals to make management decisions. We review underpinning methods A Rothman PhD,
and technologies and explore the evidence supporting the use of these devices as diagnostic and monitoring tools in Prof P V Lawford PhD,
Prof D R Hose PhD,
hypertension, arrhythmia, heart failure, coronary artery disease, pulmonary hypertension, and valvular heart disease. Prof T J A Chico MD,
Used correctly, they might improve health care and support research. Prof J P Gunn MD,
P D Morris PhD), and Insigneo
Introduction networks that can augment measurement accuracy Institute for In Silico Medicine,
(Prof P V Lawford, Prof D R Hose,
To predict, prevent, diagnose, and treat cardiovascular (eg, heart rate) or compute parameters indirectly by Prof T J A Chico, Prof J P Gunn,
diseases, physicians require an assessment of combining different signals (eg, blood pressure). WiFi P D Morris), University of
symptoms, activity, comorbidity, and context before connectivity permits near-constant data upload to cloud Sheffield, Sheffield, UK; Faculty
prescribing targeted investigations and recommen­ storage, enabling continuous and simultaneous of Medicine, Department of
Cardiology, KU Leuven, Leuven,
dations for treatment. The patient also needs monitoring of multiple parameters, which can alert Belgium
information on what the physician considers their state clinicians to significant changes (figure 2). A summary (Prof F E Rademakers MD);
of health to be, the basis for this view, and the risks and of relevant wearable sensor technologies is shown in the Dipartimento di Informatica,
benefits of any treatment. Here we examine the table. Software apps are designed to process, curate, and Università di Torino, Turin, Italy
(Prof F Ciravegna PhD);
potential for wearable technologies to enhance care in present data from raw sensor signals into relevant Department of Biomedical
all these areas of cardiovascular medicine. We review information within a user-friendly graphical display, Engineering, Eindhoven
devices that are accessible to the public, the underlying usually on a smartphone, smart watch, or linked to a University of Technology,
sensor technologies, the data acquired, and their personal computer. These data can be presented in Eindhoven, Netherlands
(Prof F N van de Vosse PhD);
application, providing a perspective on where these familiar formats such as resting heart rate and daily National Heart and Lung
tools could sit within cardiovascular health care, the step count, with multiple sensor signals combined by Institute, Faculty of Medicine,
challenges that need to be resolved, and the studies the app and displayed in device-specific formats (eg, the Imperial College London,
London, UK (Prof A Lawrie PhD,
required to confirm their utility. We also discuss the Apple Move Ring, the Garmin Fitness Age, or the Fitbit
Prof M R Wilkins MD); Academic
incorporation of sensor technologies into wearable Sleep Score). Apps can also provide user functions, such Directorate of Cardiothoracic
clothing, apparel, and cutaneous patches and how these as providing health advice based on physical activity Services, Sheffield Teaching
have been applied to clinical research in cardiovascular Hospitals NHS Foundation
Trust, Sheffield, UK (A Rothman,
medicine.
Prof T J A Chico, Prof J P Gunn,
Examples of wearable apparel Clinically useful PPG metrics
with embedded PPG sensors P D Morris); Department of
Devices, data, and apps Medicine, Stanford University,
• Heart rate monitoring
Wearable medical devices such as ambulatory blood Eyeglasses
• Arrhythmia detection
Stanford, CA, US
Hat (Prof E A Ashley Dphil);
pressure and Holter (GEHealthCare, Chicago, IL, US) Earring or earbud • Arterial oxygen saturations
Department of Circulation and
electrocardiogram (ECG) monitors, have been used in Headband • Respiratory rate
• Blood pressure Medical Imaging, Norwegian
health care for decades; nowadays, devices purchased by Shirt • Sleep assessment University of Science and
the public, including smartphones, wristbands, watches, Sports bra • Maximal oxygen consumption Technology, Trondheim,
• Pulse rate variability Norway (Prof S W Omholt PhD,
scales, shirts, rings, and eyeglasses are equipped with Armband • Arterial stiffness Prof U Wisløff PhD); School of
this type of functionality (figure 1). Patients instinctively Human Movement & Nutrition
Bracelet
recognise the potential of these technologies, and often Watch Sciences, University of
present their doctor with health app-derived data. Wristband Queensland, QLD, Australia
(Prof U Wisløff); BHF Data
Although doctors recognise their potential usefulness, Ring
Centre, Health Data Research
they are uncertain about the evidence base and UK, London, UK (Prof T J A Chico)
Sock
appropriate use of such data. Measurements range from Correspondence to:
established parameters such as heart rate, blood Dr Paul D Morris, Department of
pressure, and oxygen saturation, to step counts, minutes Figure 1: The use of PPG in wearable apparel and its application to Infection, Immunity and
cardiovascular health assessments Cardiovascular Disease,
of activity, heart rate variability, and intrathoracic PPG has been adapted to a variety of wearable apparel in published literature University of Sheffield,
impedance. Wireless technology allows multiple sensors (left panel) and how signals generated from this sensor can be applied to Sheffield S102RX, UK
to integrate different signals, enabling body area sensor cardiovascular health assessments (right panel). PPG=photoplethysmography [email protected]

www.thelancet.com/digital-health Vol 5 July 2023 e467

 Review

levels or forwarding relevant sensor data directly to Symptoms

health-care providers. Symptom context (duration, frequency, and association
All devices that claim to provide measurements of with physical activity, eating, sleeping stress, etc) is
human physiology need to show safety, accuracy, crucial to medical decision making. Traditionally, this
reliability, and reproducibility in a variety of settings. An information is obtained by asking patients to recall what
example of this is blood pressure devices, for which they were doing when they had symptoms. This
stringent, overarching guidelines are offered by information is subjective and might be misleading,
standards organisations such as the International especially for patients with memory or communication
Organization for Standardization.1 These validating difficulties. Mobile and wearable devices allow the user to
bodies approve and regulate the use of such devices to record symptoms and physical parameters simul­
ensure that health app-derived data can be relied on by taneously and has been successfully applied to mental
patients and health-care providers to accurately identify health and Parkinson’s disease.2,3 The MyHeart Counts
or exclude cardiovascular disorders. Here, there is a App study showed the feasibility of large-scale physical
specific need for greater involvement of clinical bodies activity measurement through self-reported ques­
in the development of regulatory guidelines and tionnaires, and smartphone and wearable movement
evaluation methods. data from accelerometer and gyroscope sensors with a
global positioning system (GPS). This study recruited
48 968 participants and discovered relationships between
patterns of physical activity and self-reported health.4 The
authors did, however, observe recording fatigue, limiting
2 3 long-term effectiveness. Nevertheless, there is potential
to discriminate cardiac from non-cardiac symptoms and
identify patients warranting further assessment,
investigations, or monitoring.
1

Activity
Cardiovascular diseases such as heart failure and angina
are classified according to the level of physical activity a
person reports; this, in turn, drives important treatment
decisions. Although useful, this approach is subjective
Figure 2: ECG data transmission from wearable technology and imprecise. Wearable device data provide objective
ECG data are sent via Bluetooth to a device with internet connectivity (eg, a mobile phone; step 1). Data are
transferred to the cloud storage database from the device via internet connectivity (step 2). A local platform
measurements over long periods of time. Physical
connects to the cloud storage database and the ECG recordings are reviewed by a health-care provider. activity can be quantified by accelerometers to measure
ECG=electrocardiogram parameters such as step count, or more sophisticated

Sensor principle Applications Wearable device

Accelerometer Force sensor to measure the acceleration in a single or Movement classification and recording: Smartphone, wristwatch,
multiple directions step counting, physical activity, and fall wrist band, armband, or belt
detection
Ballistocardiography Measures the body’s mechanical recoil from ventricular Cardiac energetics and cuffless blood Wristwatch or wristband
contraction to quantify cardiac outputs and dynamics pressure measurement
Electrocardiogram Measurement of electrical activity of cardiac impulse Heart rate monitoring, and heart Wristwatch or shirt or vest
rhythm assessment
Impedance- Detect changes in blood electrical conductivity to Cuffless blood pressure measurement, Wristwatch, wristband,
plethysmography measure cardiovascular characteristics (radial pulse and measurement of intrathoracic shirt, or vest
(bioimpedance) volume, and blood volume in ascending aorta) impedance in heart failure
Magnetoplethysmography Hall effect sensors track cardiovascular activity via Cuffless blood pressure measurement Wristwatch or wristband
or giant magnetoelastic movement of magnets, and giant magnetoresistance
effect sensors use the diamagnetic property of water to detect
magnetic flux created by pulsatile blood flow
Phonocardiogram Detects subaudible vibrations created through the Heart rate Patch
opening and closing of heart valves
Photoplethysmography Optical sensor to detect blood volume changes within Heart rate and rhythm, blood oxygen Wristwatch, wrist band, or
the microvascular structure of perfused tissue saturation, VO2 max, and cuffless blood eyeglasses
pressure measurement
Remote dielectric sensing Quantifies the dielectric coefficient of tissues, which Detection of pulmonary oedema Shirt or vest
correlates with fluid concentration

Table: Summary of the sensor technologies used in wearable technologies in cardiovascular disease research

e468 www.thelancet.com/digital-health Vol 5 July 2023

 Review

GPS-based measurements of speed, location, and PPG

terrain tracking over weeks or months. Ready access to
A B
such detailed and objective data at a patient-specific
Light source Light source
level will allow for a more personalised approach to Photodetector
medicine in terms of individually tailored exercise
advice, as well as monitoring response to treatment,
which is consistent with the concept of precision
medicine.
Photodetector

Heart rate C D
The relationship between heart rate and cardiovascular
risk is well established, with long-term studies showing
an association between a higher resting heart rate and
cardiovascular disease and adverse clinical outcomes.5
Heart rate monitors using chest strap electrodes have
been used in sports for decades. Photoplethysmography-
based measurements from wrist-worn devices provide E PAT
continuous monitoring with greater convenience.
Although individual device accuracy varies, recent
devices typically utilise peak detection algorithms which
report median error ranges of less than 5% in
comparison with telemetry in healthy individuals.6
Improvements in photoplethysmography sensor
ECG recording
miniaturisation has enabled incorporation into other
common apparel, including rings and eyeglasses
(figure 3).7,8 Peak detection algorithms can be confounded
by irregular rhythms—including atrial fibrillation and
premature atrial or ventricular contractions—which
produce photo­plethysmography signals that vary from Finger PPG
the underlying electrical activity. This can lead to waveform
incorrect measurements of heart rate. Sophisticated
algorithms have been developed to reduce the errors of
photoplethysmography signals in this context.9 Modern
smartwatches can offer ECG recording functions
F PTT
alongside limited rate and rhythm analysis software.10
The development of self-powered wearable devices using
triboelectric nano­ generator technology with pulse-
sensing capabilities also shows the potential for
uninterrupted heart rate monitoring over prolonged
periods.11 By harvesting the biomechanical energy Wrist PPG waveform

released by the radial artery pulse in the wrist, the self-

sustaining sensor converts the mechanical movement
into energy as well as a signal with a strong agreement
with ECG (R² 0·98).
Finger PPG waveform

Respiratory function
Commonly used respiratory parameters can also be
measured by wearable devices. Photoplethysmography
can be used to measure oxygen saturation. Many fitness Figure 3: PPG sensor principles
The intensity of either transmitted (A) or reflected (B) non-absorbed light reflects variations in blood volume.
devices also estimate maximal oxygen consumption A range of cardiovascular metrics can be calculated from the dynamic signal intensity (C). Analysis of derivatives of
during exercise (VO2 max). Respiratory rate can be this signal (D) yield additional information about blood acceleration, pressure, and arterial stiffness. PPG signal
measured with smart clothing using accelerometer, analysis can be combined with other sensor modalities to generate parameters such as PAT (the time interval
gyroscopic, or magnetometer-based detection of chest between the R wave and the PPG waveform) (E) and PTT (the time interval between the troughs of the proximal
distal PPG waveforms) (F), which can be used to estimate blood pressure. ECG=electrocardiogram.
wall movements, circumference, or impedance PPG=photoplethysmography. PAT=pulse arrival time. PTT=pulse transit time.
pneumography.12 There has been a significant rise in the
use of wearable devices to assess respiratory function
since the COVID-19 pandemic.13

www.thelancet.com/digital-health Vol 5 July 2023 e469

 Review

Ultrasonography less well against simultaneous ambulatory blood pressure

Ultrasound is central to cardiovascular assessment in monitoring outside the office (3·2 mm Hg [SD 17]).23 More
secondary care but is not widely available in primary recently, Huawei have released an inflatable cuff-based
care. Patch-based ultrasound devices now enable blood pressure wristwatch that has also shown sufficient
continuous haemodynamic measurements of arterial accuracy to meet the Association for the Advancement
Doppler waveforms and can detect changes in stroke of Medical Instrumentation, European Society of
volume.14 Obtaining such data in the community without Hypertension, and International Organization for
the need for constant specialist oversight and interval Standardization universal standards (differences between
hospital visits might enhance early detection of test and reference systolic and diastolic blood pressure
haemodynamic deterioration, particularly in valvular and –1·4 mm Hg [SD 6·47] and –0·2 mm Hg [SD 5·85],
ventricular disease. respectively);24 however, the manufacturer states that the
data obtained by the device should not be used for medical
Promoting physical activity to reduce research, diagnostic, or treatment purposes and are for
cardiovascular disease reference use only. Both products are listed as validated
Physical inactivity is considered the fourth leading risk blood pressure devices by Science and Technology for
factor for global mortality, associated with 6% of deaths Regional Innovation and Development in Europe, an
and 30% of ischaemic heart disease worldwide.15 In international organisation associated with the European
patients with symptomatic disease, activity improves Society of Hypertension, International Society of
event-free survival and might delay disease progression.16 Hypertension and World Hypertension League that aims
Despite these messages, global estimates suggest more to improve the accuracy of hypertension assessment,
than 25% of people do not achieve recommended activity diagnosis, and management by providing guidance and
levels.17 Multiple smartphone apps have been developed practice tools along with the latest scientific and validation
to reinforce healthy behaviours. One example is Active 10, evidence.25 The Aktiia bracelet (Aktiia, Neuchâtel,
which records activity levels.18 Analysis of 129 010 users Switzerland) is a cuffless blood pressure device that derives
showed a significant improvement in physical activity, measurements from an optical photo­ plethy­
smo­ graphy
including those with a very low initial baseline activity sensor. It requires an initial calibration step against a
levels. There are many other similar initiatives, including separate sphygmomanometer device, and this must be
the Smart Walk and MyHeart Counts apps. repeated intermittently. Given that there are currently no
Wearable technology lends itself to personalised guidelines for the validation of cuffless blood pressure
feedback and guidance. The Personal Activity Intelligence devices, the Aktiia bracelet was validated against an adapted
(PAI) system personalises activity targets using maximal international standard (difference between test and
heart rate data, questionnaires, and clinical data, as reference systolic and diastolic blood pressure 0·46 mm Hg
opposed to standard generic targets (10 mins brisk [SD 7·75] and 0·39 mm Hg [SD 6·86], respectively).26
walking, 10 000 steps, etc).19 In 39 298 participants, These developments highlight the willingness of the
obtaining 100 PAI points per week was associated with a commercial sector to generate and make available
23% reduction in cardiovascular disease mortality in validation data in response to calls for better regulation.27
women and 17% in men when compared with those Although these wearable devices might be more
achieving zero PAI points per week.19 A systematic review comfortable and practical for home use, and are likely to
of studies investigating the effect of personalised mobile avoid white coat hypertension due to the stress of the
health interventions in older adults showed the potential clinical environment, they do present new challenges for
to improve step count and sedentary time, although none validation; body position (particularly the position of the
achieved statistical significance.20 wrist and arm relative to the heart) and activity have been
shown to influence the blood pressure measurement.23,28
Hypertension Because of this, and the fact that these devices are
Hypertension is the leading preventable cause of death validated against traditional office-based measurements,
worldwide.21 Manufacturers have sought to incorporate the instructions for use advise only recording blood
blood pressure measurements into commercially available pressure while in an upright, seated, relaxed position.
wearable devices. Several sensor modalities have been This is an important point for physicians to consider
assessed and developed in preclinical and clinical stages when interpreting blood pressure recordings from
(table). Devices are now available to the public with the wearable devices. As the use of these devices expands it
purpose of providing validated measurements of blood might be beneficial for international organisations to
pressure from the wrist. The Omron HeartGuide, the first develop and adapt standards to consider validation of
clinically validated wristwatch blood pressure device, uses wearable sensors in the ambulant patient.
an inflation cuff within the strap to measure blood pressure
(Omron Healthcare, Kyoto, Japan).22 It compared well with Arrhythmia
office-based sphygmomanometer measurements (diff­ Prolonged ECG monitoring devices such as Holter
erence in systolic blood pressure <5mm Hg [SD 8]), but monitors are cumbersome, can only be worn for short

e470 www.thelancet.com/digital-health Vol 5 July 2023

 Review

periods, and might miss paroxysmal arrhythmias. ring (Oura Health, Oulu, Finland), but it has only been
Furthermore, implantable devices (loop recorders) are tested in sinus rhythm to date.7 A variety of wearable
invasive and require specialist training to insert and designs might soon become available for consumer-driven
analyse. Wearable technology can be advantageous for arrhythmia monitoring, allowing device selection that
arrhythmia detection; several modern smartwatches use best suits patient preference and comfort.
photo­plethysmography sensors, and more recently ECG
technology, to detect heart rate and rhythm. Photo­ Heart failure
plethysmography sensors can detect atrial fibrillation, the In the management of congestive heart failure (CHF),
most common significant arrhythmia, with a sensitivity reduced activity levels are predictive of worse outcomes,
and specificity of 91–100% in comparison with ECG.29,30 including mortality.36 Wearable actigraphy devices have
The Apple Heart study investigated the use of smartwatch- been validated against patient diaries and physical activity
based arrhythmia detection in 419 297 participants (Apple, during cardiac rehabilitation.37 Wrist-based actigraphy
Cupertino, CA, USA).31 Through the watch’s photo­ enabled the dichotomisation of 50 patients with CHF into
plethysmography sensor, if an irregular cardiac rhythm higher and lower physical activity groups, with the lower
was detected, a notification advised a telemedicine physical activity group having a five times higher rate of
consultation and to wear an ECG recording patch for hospital admission.38 A systematic review of wearable
7 days. Within a median monitoring period of 117 days, actigraphy monitoring in patients with CHF revealed
2 161 participants received an irregular pulse notifi­ reduced physical activity was associated with poor clinical
cation (0·52%). After exclusions, 450 participants outcomes, mortality, and morbidity.39 The AWAKE-HF
returned usable ECG data, 34% of which were confirmed study, which assessed the effect of combination sacubitril
to have atrial fibrillation. Although undergoing simul­ and valsartan versus enalapril on quality of life in patients
taneous ECG monitoring, the positive predictive value of with heart failure with reduced ejection fraction, used
subsequent irregular pulse notifi­cations was 84%. In a wearable actigraphy to detect physical activity. The study
similar study performed across China, 246 541 people showed no significant difference in activity between either
downloaded a mobile atrial fibrillation app that used data treatment group, despite an improvement in patient-
from a photoplethysmography wristband or watch. Of reported quality of life in the group who received sacubitril
these, 187 912 people used the app via their smartphone and valsartan, suggesting a divergence between objectively
and 0·23% received a notification of a suspected atrial gathered patient data and subjective outcomes.40 A band
fibrillation. After follow-up, the positive predictive value electrode design integrated in shirts or vests can measure
of the photoplethysmography detected signal was 91·6%.32 and detect changes in intrathoracic impedance which
These studies were the first demonstration of a general correlates with weight reduction secondary to diuresis
population-wide approach to arrhythmia screening using (figure 4). Monitoring this parameter might predict
a commercially available wearable device with tele­ hospital admission more reliably than change in weight
medicine. Given that atrial fibrillation was diagnosed in alone.41,42 One study investigated heart failure patients
asymptomatic participants, future research should wearing such a vest for only 5 min a day after discharge
consider whether major sequelae such as the incidence of from hospital, with data transmitted via Bluetooth and
stroke can be reduced by this approach in adequately mobile phone.43 For 106 participants, there were 64 heart
powered prospective controlled trials. failure events (18 readmissions and 46 up-titrations of
Heart rate-sensing wearable technology has also been diuretic medication). An algorithm analysing intrathoracic
applied to multiple types of apparel, leading to a variety of impedance was consequently developed, which showed
potential arrhythmia detection modalities. The smart shirt 87% sensitivity and 70% specificity in identifying patients
design Cardioskin (BioSerenity, Paris, France) can record with recurrent admissions due to decompensated heart
a 15 lead ECG continuously with quality comparable to failure. Remote dielectric sensing (ReDS) uses electro­
Holter monitors.33 ECG vests or chest straps worn for magnetic signals emitted across the chest. It correlates
28 days were investigated in the assessment of 146 patients well with pulmonary congestion and has been developed
with a diagnosis of cryptogenic stroke.34 Atrial fibrillation for home monitoring and wearable apps.44,45 Daily ReDS
was detected in 21·9% of patients, with the number monitoring in 50 patients with CHF for 90 days, with
needed to screen to detect one incidence of atrial appropriate medication changes, resulted in an
fibrillation being 4·8 patients. Smart eyewear might also 87% reduction in admissions compared with the pre-
provide additional opportunities. A smartphone camera- monitoring period, and 79% reduction compared with
based program can detect atrial fibrillation through facial the post-monitoring period.46 Retrospective studies
photo­plethysmography using variations detected in skin corroborate these findings.47,48
colour, with a sensitivity of 95% and specificity of 96% in
discriminating atrial fibrillation from sinus rhythm.35 This Ischaemic heart disease
introduces the potential for facial photo­plethysmography Despite the evidence supporting physical activity and
to be developed for eyeglasses.8 Miniaturisation of cardiac rehabilitation after acute coronary syndrome (ACS)
photoplethysmography has led to a heart rate monitoring and revascularisation, studies using activity-monitoring

www.thelancet.com/digital-health Vol 5 July 2023 e471

 Review

l(t)
quality-of-life questionnaires).56–58 This does not always
correlate with right heart catheter measurements and
echocardiographic changes, suggesting that fatigue and
reduced physical activity might be the result of a complex
multi-system response.59 The FIT-PH study aims to
evaluate the correlation of Fitbit wristband activity data
with implanted pulmonary artery pressure monitoring
devices and their relationship to the patient’s clinical
condition and quality of life (NCT04078243). When used
to monitor response to a physical training intervention, a
study using wrist-based accelerometers showed an
increase in physical activity in the intervention group,
along with an improvement in 6 min walk distance
(6MWD).60 Using armband accelerometers, inspiratory
U(t) muscle training failed to show an increase in daily physical
Figure 4: Intrathoracic impedance measurement technique
activity levels, 6MWD, or questionnaire responses against
Electrodes on both sides of the chest (green and blue) pass a constant current controls.61 Drug and device trials in pulmonary hyper­
across the chest, I(t), with an impedance detection circuit (yellow) measuring tension have incorporated activity monitoring as
the voltage drop caused by intrathoracic blood volume U(t). outcomes. Inhaled nitric oxide therapy was assessed in a
randomised, double-blind, placebo-controlled trial.62
devices report daytime sedentary periods in excess of 9 h Using armband accelero­meters, 23% of the people in the
within the first month of recovery.49,50 A prospective study treatment group showed a significant improvement in
of 72 patients with stable ischaemic heart disease taking physical activity following an 8 week treatment period,
optimal medical therapy demonstrated an improvement in whereas no improvement was observed in the placebo
exercise time following a 6-week cardiac rehabilitation group; 71% of people in the placebo group had a significant
programme over a 12 month follow-up period; however, decrease in physical activity versus 39% in the treatment
the proportion of patients who met guideline group.62 The TROPHY1 feasibility study, investigating the
recommendations of physical activity targets did not use of pulmonary artery denervation, showed an
change significantly and sedentary time remained high improvement in physical activity monitored through
throughout.51 Patients with the lowest level of sedentary wearable accelero­metery alongside conventional metrics.63
time made the most improvement in physical activity, with Other recent trials in pulmonary hypertension such as
the reverse also being true. A study of 330 patients who VENTASTEP,64 which investigated inhaled iloprost, and
provided wristwatch-based accelerometer data following TRACE,65 which investigated selective prostacyclin
discharge after an ACS revealed that only 16% conformed receptor therapy, used wearable technology to measure
to exercise guidelines.52 Simply wearing an activity monitor parameters of daily physical activity as primary outcome
might encourage physical activity. Small randomised measures.64,65
controlled studies of patients attending cardiac rehab­
ilitation demonstrate that wearable pedometers improved Aortic valve disease
adherence with physical activity advice, with improvements Data from wearable devices are surprisingly sparse in
in psychosocial health and self-reported function;53,54 aortic valve disease, considering their potential to track
however, these studies also included support from progression and optimally time intervention. In a study
specialist staff which might have contributed to the results. of 52 patients with severe aortic stenosis, there was little
The UP-STEP ACS study utilising the Fitbit Charge 2 correlation between wrist-based accelerometer measure­
(Fitbit, San Francisco, CA, USA) will be the first to assess ments of daytime physical activity and conventional
the effect of wearable physical activity monitoring in performance assessments and self-reported activity
improving exercise capacity and modifying cardiovascular questionnaires.66 Wrist-mounted accelerometer-based
risk factors in a randomised setting in recovering patients daily activity monitoring in 25 patients undergoing
with ACS.55 transcatheter aortic valve implantation showed a recovery
to pre-transcatheter aortic valve implantation levels of
Pulmonary hypertension physical activity by 5 weeks.67 Further studies are required
Pulmonary hypertension is a debilitating condition with to explore the relationship between activity level and
significant morbidity and mortality, and high levels of other physiological biomarkers with prognosis and
fatigue and reduced physical activity. Accelerometer-based intervention.
sensors detected reduced daily step counts, distance
walked, and time spent in moderate to vigorous physical Computational and predictive modelling
activity, with good correlations with established clinical By combining the various sources of information from
parameters of physical activity (eg, 6 min walk test and wearable technology into personal representations,

e472 www.thelancet.com/digital-health Vol 5 July 2023

 Review

computer modelling could substantially enhance the Handling data from wearable technology in the
information that can be extracted from such data.68 clinic
Modern medicine is guided by randomised controlled Incorporating data from wearable devices into clinical
trials, which report relatively small overall effect sizes in decision making appears intuitive, plausible, and
large, heterogeneous patient groups. The results are attractive. Patients are already presenting to their doctors
then applied, often by extrapolation, as a generic with data from their wearable device as a new, patient-
remedy. Even if an individual patient in the clinic would driven initiative. The collection of relevant health data
have met the inclusion criteria of the key randomised before the first clinical consultation has the potential to
control trial, would they have been one of the few revolutionise the traditional doctor–patient interaction
positive responders? The individual’s personal circum­ (figure 5). Patients might be anxious about what they
stances, comorbidities, socioeconomic status, genetics, consider abnormal data, even without symptoms, and
activity levels, ethnicity, age, frailty, and other factors might (paradoxically) curtail their exercise as a result. At
might be relevant. Computational modelling can this early stage, doctors should ensure that data have
analyse complex datasets to determine relevant factors been obtained within the manufacturer’s stated intended
and associations, offering the goal of tailored, or use and do not exceed its evidence base or breach its
personalised care, rather than a one-size-fits-all regulatory approval. Off-label use might transfer
approach.69 Wearable technology has the potential to responsibility from the manufacturer to the end user.
obtain patient-specific data to tune these models in an
individualised manner. Continuous assimilation of Challenges and future perspectives
physiological data over prolonged periods will allow The ubiquity of wearable devices risks that they will be
computational models to develop and age with the used in clinical decision making, irrespective of
patient; the so called digital twin.70 It is anticipated that validation. Despite a wide range of studies, they have not
such models will incorporate artificial intelligence yet been shown to improve defined medical outcomes,
technology to characterise and delineate patient-specific necessitating studies to identify risks and benefits to the
physiological relationships that will be used to predict patient and health-care systems. Medical training will
the likelihood of a new diagnosis, disease decom­ need to address these risks and benefits as the evidence
pensation, or response to an intervention.71 In the base grows, as well as navigate the difficulties posed by
All of Us research programme, which includes more volumes of disparate device data. Cardiologists, who are
than 400 000 participants, data from wearable devices familiar with assessing traditional cardiovascular
are combined with surveys, electronic health records, symptoms such as chest pain, palpitations, and shortness
and clinical and laboratory samples to advance precision of breath, based on a patient’s subjective description, will
diagnosis on a large scale.72 need to assimilate these new data sources and, in time,

Figure 5: How wearable devices can disrupt the traditional doctor–patient interaction
(A) The traditional model of healthcare is initiated by symptoms, leading to a patient–clinician consultation. The clinician then requests investigations that document
objective cardiovascular parameters under resting and stress conditions that leads to reassurance or diagnosis and treatment. (B) Patients present to their doctor with
objective cardiovascular measurements under resting and physical exercise conditions, as measured and alerted by their wearable device. This process might lead to
earlier initiation of treatment or reassurance and might facilitate monitoring the response to treatment.

www.thelancet.com/digital-health Vol 5 July 2023 e473

 Review

PDM and GJW designed the figures used. All authors contributed to
Search strategy and selection criteria subsequent drafting and editing of the manuscript.

The Web of Science and PubMed databases were used to Declaration of interests
EAA is an advisor to Apple and has received funding from Apple,
search for relevant journal articles, including clinical trials,
Google, and Samsung. AL receives research support from Apple.
meta-analyses, and randomised controlled trials. The search PDM has received speaker’s honoraria from Abbot. All other authors
terms used were “wearable”, “sensors”, “cardiovascular”, declare no competing interests.
“blood pressure”, “cuffless”, “heart disease”, and “physical Acknowledgments
activity”. Publications accessible from the database inception All figures created with Biorender.com. This work was supported by the
up to Nov 1, 2022 were considered for inclusion based on National Institute for Health and Care Research (NIHR) Sheffield
Biomedical Research Centre (BRC). The views expressed are those of the
their relevance to consumer-led wearable devices in authors and not necessarily those of the NIHR or the Department of
cardiovascular medicine. Articles were not included or Health and Social Care. PDM was funded by the Wellcome Trust (grant
excluded according to language. number 214567/Z/18/Z). GJW was funded by the Engineering and
Physical Sciences Research Council (scholarship number 199860907).
AA-B was supported by King Saud bin Abdulaziz University for Health
Sciences via the Saudi Cultural Bureau. AL was supported by the British
use them to monitor the response to treatment. Parallel Heart Foundation (grant number FS/18/52/33808). AR was funded by
developments in machine learning and artificial the Wellcome Trust (grant number 206632/Z/17/z), MRC Experimental
intelligence might also be applied to data from wearable Medicine Award (grant number MR/W026279/1), Medtronic External
devices to identify novel associations in terms of Research Program, Abbott Investigator Sponsor Studies Award,
Endotronix, Novartis, and Janssen. MRW was supported by the British
diagnosis, risk prediction, and treatment choices.71 It is Heart Foundation (grant number RE/18/4/34215) and the National
also likely that sensor technology will advance beyond Institute for Health and Care Research. TJAC was supported by the
traditional methods such as photoplethysmography with Engineering and Physical Sciences Research Council. None of the
funding sources had any role in the writing of the manuscript, and none
the emergence of smart fabrics. These incorporate novel
of the authors have been paid to write this manuscript.
sensing methods, including soft magnetoelastic
References
generators and hierarchical in situ filling porous 1 International Organization for Standardisation. ISO 81060-2:2018.
piezoresistive sensors capable of transducing pressure Non-invasive sphygmomanometers—Part 2: Clinical investigation
displacement. Such fabrics provide new ways of of intermittent automated measurement type. Geneva:
International Organization for Standardisation, 2018.
integrating biomedical sensing technology into clothing 2 Rohani DA, Faurholt-Jepsen M, Kessing LV, Bardram JE.
and wearable devices.73,74 The regulatory landscape for Correlations between objective behavioral features collected from
digital health care will need to develop rapidly. Public mobile and wearable devices and depressive mood symptoms in
patients with affective disorders: systematic review.
trust and acceptance will also be required. One hope is JMIR Mhealth Uhealth 2018; 6: e165.
that wearables might help to reduce inequalities in health 3 Bot BM, Suver C, Neto EC, et al. The mPower study, Parkinson
care, provided they are inexpensive, and do not disease mobile data collected using ResearchKit. Sci Data 2016;
3: 160011.
discriminate against individuals who may struggle to
4 Hershman SG, Bot BM, Shcherbina A, et al. Physical activity, sleep
engage with the technology (eg, due to a lack of familiarity and cardiovascular health data for 50,000 individuals from the
with electronic devices or health conditions which may MyHeart Counts Study. Sci Data 2019; 6: 24.
make using small devices challenging). The combination 5 Jensen MT. Resting heart rate and relation to disease and longevity:
past, present and future. Scand J Clin Lab Invest 2019; 79: 108–16.
of wearables with telemedicine might also lead to a 6 Shcherbina A, Mattsson CM, Waggott D, et al. Accuracy in wrist-
revolution in community care, as well as a reduction in worn, sensor-based measurements of heart rate and energy
both acute hospital admission and health spending. expenditure in a diverse cohort. J Pers Med 2017; 7: 3.
7 Kinnunen HO, Koskimäki H. 0312 The HRV Of The Ring -
comparison of nocturnal HR and HRV between a commercially
Conclusion available wearable ring and ECG. Sleep 2018; 41 (suppl 1): A120.
Wearable technology has the potential to elevate the 8 Howell TA, Chao D, Thomas CD, Day RG, Tong PP. Eyeglasses
routine clinical consultation from a subjective discussion with a heart rate monitor. US Patent No. 7677723B2. Washington
DC: US Patent and Trademark Office, 2010.
based on patient recollection to a standardised series of 9 Han D, Bashar SK, Lázaro J, et al. A real-time PPG peak detection
objective parameters of both health and quality of life method for accurate determination of heart rate during sinus
gathered over months or years. Wearable devices have rhythm and cardiac arrhythmia. Biosensors 2022; 12: 82.
10 Apple Support. Taking an ECG with the ECG app on the Apple
shown some potential application in many cardiovascular Watch. Sept 12, 2022. https://support.apple.com/en-us/HT208955
diseases, either through disease screening or monitoring; (accessed Sept 15, 2022).
however, the evidence base and the integration of 11 Ouyang H, Tian J, Sun G, et al. Self-powered pulse sensor for
antidiastole of cardiovascular disease. Adv Mater 2017; 29: 1703456.
wearable data into clinical cardiology is still in its infancy.
12 Aliverti A. Wearable technology: role in respiratory health and
Such data could soon help personalise and improve the disease. Breathe 2017; 13: e27–36.
management of cardiovascular diseases on multiple 13 Takahashi S, Nakazawa E, Ichinohe S, Akabayashi A,
levels, ultimately resulting in better outcomes on both an Akabayashi A. Wearable technology for monitoring respiratory rate
and spo2 of covid-19 patients: a systematic review. Diagnostics 2022;
individual and a population-wide scale. 12: 2563.
Contributors 14 Wang C, Li X, Hu H, et al. Monitoring of the central blood pressure
PDM, GJW, JPG, TJAC, and DRH conceived the idea for this Review. waveform via a conformal ultrasonic device. Nat Biomed Eng 2018;
PDM, GJW, and JPG contributed to the original manuscript draft. 2: 687–95.

e474 www.thelancet.com/digital-health Vol 5 July 2023

 Review

15 Kenny JS, Clarke G, Myers M, et al. A wireless wearable doppler 35 Yan BP, Lai WHS, Chan CKY, et al. Contact-free screening of atrial
ultrasound detects changing stroke volume: proof-of-principle fibrillation by a smartphone using facial pulsatile
comparison with trans-esophageal echocardiography during photoplethysmographic signals. J Am Heart Assoc 2018;
coronary bypass surgery. Bioengineering 2021; 8: 203. 7: e008585.
16 WHO. Global health risks: mortality and burden of disease 36 Walsh JT, Charlesworth A, Andrews R, Hawkins M, Cowley AJ.
attributable to selected major risks. Geneva: World Health Relation of daily activity levels in patients with chronic heart failure
Organization, 2009. to long-term prognosis. Am J Cardiol 1997; 79: 1364–69.
17 Winzer EB, Woitek F, Linke A. Physical activity in the prevention 37 Deka P, Pozehl B, Norman JF, Khazanchi D. Feasibility of using the
and treatment of coronary artery disease. J Am Heart Assoc 2018; Fitbit® Charge HR in validating self-reported exercise diaries in a
7: e007725. community setting in patients with heart failure.
18 Ciravegna F, Gao J, Ireson N, Copeland R, Walsh J, Lanfranchi V. Eur J Cardiovasc Nurs 2018; 17: 605–11.
Active 10: Brisk walking to support regular physical activity. In: 38 Waring T, Gross K, Soucier R, ZuWallack R. Measured physical
PervasiveHealth. Proceedings of 13th EAI International Conference on activity and 30-day rehospitalization in heart failure patients.
Pervasive Computing Technologies for Healthcare 2019; 11–20. J Cardiopulm Rehabil Prev 2017; 37: 124–29.
19 Nes BN, Gutvik CR, Lavie CL, Nauman J, Wisløff U. Personalized 39 Tan MKH, Wong JKL, Bakrania K, et al. Can activity monitors
activity intelligence (PAI) for prevention of cardiovascular disease predict outcomes in patients with heart failure? A systematic review.
and promotion of physical activity. Am J Med 2017; 130: 328–36. Eur Heart J Qual Care Clin Outcomes 2019; 5: 11–21.
20 Broers ER, Kop WJ, Denollet J, et al. A personalized ehealth 40 Khandwalla RM, Grant D, Birkeland K, et al. The AWAKE-HF
intervention for lifestyle changes in patients with cardiovascular study: sacubitril/valsartan impact on daily physical activity and
disease: randomized controlled trial. J Med Internet Res 2020; sleep in heart failure. Am J Cardiovasc Drugs 2021; 21: 241–54.
22: e14570. 41 Cuba-Gyllensten I, Gastelurrutia P, Riistama J, et al. A novel
21 Forouzanfar MH, Afshin A, Alexander LT, et al. Global, regional, wearable vest for tracking pulmonary congestion in acutely
and national comparative risk assessment of 79 behavioural, decompensated heart failure. Int J Cardiol 2014; 177: 199–201
environmental and occupational, and metabolic risks or clusters of 42 Cuba Gyllensten I, Bonomi AG, Goode KM, et al. Early indication
risks, 1990–2015: a systematic analysis for the Global Burden of of decompensated heart failure in patients on home-telemonitoring:
Disease Study 2015. Lancet 2016; 388: 1659–724. a comparison of prediction algorithms based on daily weight and
22 Kuwabara M, Harada K, Hishiki Y, Kario K. Validation of two watch- noninvasive transthoracic bio-impedance. JMIR Med Inform
type wearable blood pressure monitors according to the ANSI/ 2016; 4: e3.
AAMI/ISO81060-2:2013 guidelines: Omron HEM-6410T-ZM and 43 Darling CE, Dovancescu S, Saczynski JS, et al. Bioimpedance-based
HEM-6410T-ZL. J Clin Hypertens 2019; 21: 853–58. heart failure deterioration prediction using a prototype fluid
23 Kario K, Shimbo D, Tomitani N, Kanegae H, Schwartz JE, accumulation vest-mobile phone dyad: an observational study.
Williams B. The first study comparing a wearable watch-type blood JMIR Cardio 2017; 1: e1.
pressure monitor with a conventional ambulatory blood pressure 44 Amir O, Rappaport D, Zafrir B, Abraham WT. A novel approach to
monitor on in-office and out-of-office settings. J Clin Hypertens monitoring pulmonary congestion in heart failure: initial animal
2020; 22: 135–41. and clinical experiences using remote dielectric sensing technology.
24 Zhang W, Zhou YN, Zhou Y, Wang JG. Validation of the watch-type Congest Heart Fail 2013; 19: 149–55.
HUAWEI WATCH D oscillometric wrist blood pressure monitor in 45 Amir O, Azzam ZS, Gaspar T, et al. Validation of remote dielectric
adult Chinese. Blood Press Monit 2022; 27: 353–56. sensing (ReDS™) technology for quantification of lung fluid status:
25 STRIDE BP. Joint Initiative with ESH (European Society of comparison to high resolution chest computed tomography in
Hypertension), International Society of Hypertension and World patients with and without acute heart failure. Int J Cardiol 2016;
Hypertension League: validated blood pressure monitors. https:// 221: 841–46.
www.stridebp.org/bp-monitors (accessed Nov 20, 2022). 46 Amir O, Ben-Gal T, Weinstein JM, et al. Evaluation of remote
26 Vybornova A, Polychronopoulou E, Wurzner-Ghajarzadeh A, dielectric sensing (ReDS) technology-guided therapy for decreasing
Fallet S, Sola J, Wuerzner G. Blood pressure from the optical Aktiia heart failure re-hospitalizations. Int J Cardiol 2017; 240: 279–84.
Bracelet: a 1-month validation study using an extended ISO81060-2 47 Roy S, Zafar A, Vazquez CE, et al. Noninvasive remote dielectric
protocol adapted for a cuffless wrist device. Blood Press Monit 2021; sensing vest significantly reduces readmission rate of patients with
26: 305–11. heart failure. J Card Fail 2018; 24: S92.
27 Campbell NR, Gelfer M, Stergiou GS, et al. A call to regulate 48 Opsha Y, Zhuge P, Guevarra J, Lundberg M, Rogal G. Retrospective
manufacture and marketing of blood pressure devices and cuffs: evaluation of remote dielectric sensing (ReDS) vest technology and
a position statement from the world hypertension league, its impact on heart failure readmission rates and diuretics therapy.
international society of hypertension and supporting hypertension J Card Fail 2019; 25 (suppl): S147–48.
organizations. J Clin Hypertens 2016; 18: 378–80. 49 Freene N, Borg S, McManus M, et al. Comparison of device-based
28 Sola J, Vybornova A, Fallet S, Polychronopoulou E, physical activity and sedentary behaviour following percutaneous
Wurzner-Ghajarzadeh A, Wuerzner G. Validation of the optical coronary intervention in a cohort from Sweden and Australia:
Aktiia bracelet in different body positions for the persistent a harmonised, exploratory study. BMC Sports Sci Med Rehabil 2020;
monitoring of blood pressure. Sci Rep 2021; 11: 20644. 12: 1–8.
29 Tison GH, Sanchez JM, Ballinger B, et al. Passive detection of atrial 50 Duran AT, Ewing Garber C, Cornelius T, Schwartz JE, Diaz KM.
fibrillation using a commercially available smartwatch. Patterns of sedentary behavior in the first month after acute
JAMA Cardiol 2018; 3: 409–16. coronary syndrome. J Am Heart Assoc 2019; 8: e011585.
30 Bonomi AG, Schipper F, Eerikäinen LM, et al. Atrial fibrillation 51 Freene N, McManus M, Mair T, Tan R, Davey R. High sedentary
detection using a novel cardiac ambulatory monitor based on photo- behaviour and low physical activity levels at 12 months after cardiac
plethysmography at the wrist. J Am Heart Assoc 2018; 7: e009351. rehabilitation: a prospective cohort study. Ann Phys Rehabil Med
31 Perez MV, Mahaffey KW, Hedlin H, et al. Large-scale assessment of 2020; 63: 53–58.
a smartwatch to identify atrial fibrillation. N Engl J Med 2019; 52 Kronish IM, Diaz KM, Goldsmith J, Moise N, Schwartz JE.
381: 1909–17. Objectively measured adherence to physical activity guidelines after
32 Guo Y, Wang H, Zhang H, et al. Mobile photoplethysmographic acute coronary syndrome. J Am Coll Cardiol 2017; 69: 1205–07.
technology to detect atrial fibrillation. J Am Coll Cardiol 2019; 53 Butler L, Furber S, Phongsavan P, Mark A, Bauman A. Effects of a
74: 2365–75. pedometer-based intervention on physical activity levels after
33 Fouassier D, Roy X, Blanchard A, Hulot J-S. Assessment of signal cardiac rehabilitation: a randomized controlled trial.
quality measured with a smart 12-lead ECG acquisition T-shirt. J Cardiopulm Rehabil Prev 2009; 29: 105–14.
Ann Noninvasive Electrocardiol 2020; 25: e12682. 54 Houle J, Doyon O, Vadeboncoeur N, Turbide G, Diaz A, Poirier P.
34 Pagola J, Juega J, Francisco-Pascual J, et al. Yield of atrial fibrillation Effectiveness of a pedometer-based program using a socio-cognitive
detection with textile wearable holter from the acute phase of stroke: intervention on physical activity and quality of life in a setting of
pilot study of Crypto-AF registry. Int J Cardiol 2018; 251: 45–50. cardiac rehabilitation. Can J Cardiol 2012; 28: 27–32.

www.thelancet.com/digital-health Vol 5 July 2023 e475

 Review

55 Nogic J, Thein PM, Cameron J, Mirzaee S, Ihdayhid A, Nasis A. 65 Howard LS, Rosenkranz S, Frantz RP, et al. Assessing daily life
The utility of personal activity trackers (Fitbit Charge 2) on exercise physical activity by actigraphy in pulmonary arterial hypertension:
capacity in patients post acute coronary syndrome [UP-STEP ACS insights from the randomized controlled study with selexipag
Trial]: a randomised controlled trial protocol. BMC Cardiovasc Disord (TRACE). Chest 2023; 163: 407–18.
2017; 17: 1–9. 66 Tang Y, Green P, Maurer M, et al. Relationship between
56 González-Saiz L, Santos-Lozano A, Fiuza-Luces C, et al. Physical accelerometer-measured activity and self-reported or performance-
activity levels are low in patients with pulmonary hypertension. based function in older adults with severe aortic stenosis.
Ann Transl Med 2018; 6: 205. Curr Geriatr Rep 2015; 4: 377–84.
57 Cascino TM, McLaughlin VV, Richardson CR, et al. Physical activity 67 Green P, Woglom A, Maurer M, et al. Accelerometer-measured
and quality of life in patients with pulmonary hypertension. physical activity recovery in the first month after transcatheter aortic
Eur Respir J 2019; 53: 1900028. valve replacement. J Am Coll Cardiol 2012; 59 (suppl): E1964 (abstract).
58 Sehgal S, Chowdhury A, Rabih F, et al. Counting steps: a new way 68 King RC, Villeneuve E, White RJ, Sherratt RS, Holderbaum W,
to monitor patients with pulmonary arterial hypertension. Lung Harwin WS. Application of data fusion techniques and technologies
2019; 197: 501–08. for wearable health monitoring. Med Eng Phys 2017; 42: 1–12.
59 Saxer S, Lichtblau M, Berlier C, Hasler ED, Schwarz EI, Ulrich S. 69 Hose DR, Lawford PV, Huberts W, Hellevik LR, Omholt SW,
Physical activity in incident patients with pulmonary arterial and van de Vosse FN. Cardiovascular models for personalised medicine:
chronic thromboembolic hypertension. Lung 2019; 197: 617–25. where now and where next? Med Eng Phys 2019; 72: 38–48.
60 González-Saiz L, Fiuza-Luces C, Sanchis-Gomar F, et al. Benefits of 70 de Lepper AGW, Buck CMA, van ’t Veer M, Huberts W,
skeletal-muscle exercise training in pulmonary arterial van de Vosse FN, Dekker LRC. From evidence-based medicine to
hypertension: The WHOLEi+12 trial. Int J Cardiol 2017; 231: 277–83. digital twin technology for predicting ventricular tachycardia in
61 Aslan GK, Akıncı B, Yeldan I, Okumus G. A randomized controlled ischaemic cardiomyopathy. J R Soc Interface 2022; 19: 20220317.
trial on inspiratory muscle training in pulmonary hypertension: 71 Krittanawong C, Rogers AJ, Johnson KW, et al. Integration of novel
effects on respiratory functions, functional exercise capacity, monitoring devices with machine learning technology for scalable
physical activity, and quality of life. Heart Lung 2020; 49: 381–87. cardiovascular management. Nat Rev Cardiol 2021; 18: 75–91.
62 Nathan SD, Flaherty KR, Glassberg MK, et al. A randomized, 72 Ramirez AH, Gebo KA, Harris PA. Progress with the ALL of US
double-blind, placebo-controlled study of pulsed, inhaled nitric research program: opening access for researchers. JAMA 2021;
oxide in subjects at risk of pulmonary hypertension associated with 325: 2441–42.
pulmonary fibrosis. Chest 2020; 158: 637–45. 73 Zhou Y, Zhao X, Xu J, et al. Giant magnetoelastic effect in soft
63 Rothman AMK, Vachiery J-L, Howard LS, et al. Intravascular systems for bioelectronics. Nat Mater 2021; 20: 1670–76.
ultrasound pulmonary artery denervation to treat pulmonary 74 Xu J, Li H, Yin Y, et al. High sensitivity and broad linearity range
arterial hypertension (TROPHY1): multicenter, early feasibility pressure sensor based on hierarchical in-situ filling porous
study. JACC Cardiovasc Interv 2020; 13: 989–99. structure. npj Flex Electron 2022; 6: 62.
64 Mueller C, Stollfuss B, Roitenberg A, Harder J, Richter MJ.
Evaluation of clinical outcomes and simultaneous digital tracking of Copyright © 2023 The Author(s). Published by Elsevier Ltd. This is an
daily physical activity, heart rate, and inhalation behavior in patients Open Access article under the CC BY 4.0 license.
with pulmonary arterial hypertension treated with inhaled iloprost:
protocol for the observational VENTASTEP study. JMIR Res Protoc
2019; 8: e12144.

e476 www.thelancet.com/digital-health Vol 5 July 2023