Performance of an automated photoplethysmography-based artificial intelligence algorithm to detect atrial fibrillation

Design of the Dutch multicentre study on opportunistic screening of geriatric patients for atrial fibrillation using a smartphone PPG app: the Dutch-GERAF study

BACKGROUND

Screening of high-risk patients is advocated to achieve early detection and treatment of clinical atrial fibrillation (AF). The Dutch-GERAF study will address two major issues. Firstly, the effectiveness and feasibility of an opportunistic screening strategy for clinical AF will be assessed in frail older patients and, secondly, observational data will be gathered regarding the efficacy and safety of oral anticoagulation (OAC).

METHODS

This is a multicentre study on opportunistic screening of geriatric patients for clinical AF using a smartphone photoplethysmography (PPG) application. Inclusion criteria are age ≥ 65 years and the ability to perform at least three PPG recordings within 6 months. Exclusion criteria are the presence of a cardiac implantable device, advanced dementia or a severe tremor. The PPG application records patients' pulse at their fingertip and determines the likelihood of clinical AF. If clinical AF is suspected after a positive PPG recording, a confirmatory electrocardiogram is performed. Patients undergo a comprehensive geriatric assessment and a frailty index is calculated. Risk scores for major bleeding (MB) are applied. Standard laboratory testing and additional laboratory analyses are performed to determine the ABC-bleeding risk score. Follow-up data will be collected at 6 months, 12 months and 3 years on the incidence of AF, MB, hospitalisation, stroke, progression of cognitive disorders and mortality.

DISCUSSION

The Dutch-GERAF study will focus on frail older patients, who are underrepresented in randomised clinical trials. It will provide insight into the effectiveness of screening for clinical AF and the efficacy and safety of OAC in this high-risk population.

TRIAL REGISTRATION

NCT05337202.

Photoplethysmography based atrial fibrillation detection: a continually growing field

Objective. Atrial fibrillation (AF) is a prevalent cardiac arrhythmia associated with significant health ramifications, including an elevated susceptibility to ischemic stroke, heart disease, and heightened mortality. Photoplethysmography (PPG) has emerged as a promising technology for continuous AF monitoring for its cost-effectiveness and widespread integration into wearable devices. Our team previously conducted an exhaustive review on PPG-based AF detection before June 2019. However, since then, more advanced technologies have emerged in this field.Approach. This paper offers a comprehensive review of the latest advancements in PPG-based AF detection, utilizing digital health and artificial intelligence (AI) solutions, within the timeframe spanning from July 2019 to December 2022. Through extensive exploration of scientific databases, we have identified 57 pertinent studies.Significance. Our comprehensive review encompasses an in-depth assessment of the statistical methodologies, traditional machine learning techniques, and deep learning approaches employed in these studies. In addition, we address the challenges encountered in the domain of PPG-based AF detection. Furthermore, we maintain a dedicated website to curate the latest research in this area, with regular updates on a regular basis.

Artificial Intelligence-Based Clinical Decision Support Systems in Cardiovascular Diseases

Despite all the advancements in science, medical knowledge, healthcare, and the healthcare industry, cardiovascular disease (CVD) remains the leading cause of morbidity and mortality worldwide. The main reasons are the inadequacy of preventive health services and delays in diagnosis due to the increasing population, the failure of physicians to apply guide-based treatments, the lack of continuous patient follow-up, and the low compliance of patients with doctors' recommendations. Artificial intelligence (AI)-based clinical decision support systems (CDSSs) are systems that support complex decision-making processes by using AI techniques such as data analysis, foresight, and optimization. Artificial intelligence-based CDSSs play an important role in patient care by providing more accurate and personalized information to healthcare professionals in risk assessment, diagnosis, treatment optimization, and monitoring and early warning of CVD. These are just some examples, and the use of AI for CVD decision support systems is rapidly evolving. However, for these systems to be fully reliable and effective, they need to be trained with accurate data and carefully evaluated by medical professionals.

Accuracy of Artificial Intelligence-Based Technologies for the Diagnosis of Atrial Fibrillation: A Systematic Review and Meta-Analysis

Atrial fibrillation (AF) is the most common arrhythmia with a high burden of morbidity including impaired quality of life and increased risk of thromboembolism. Early detection and management of AF could prevent thromboembolic events. Artificial intelligence (AI)--based methods in healthcare are developing quickly and can be proved as valuable for the detection of atrial fibrillation. In this metanalysis, we aim to review the diagnostic accuracy of AI-based methods for the diagnosis of atrial fibrillation. A predetermined search strategy was applied on four databases, the PubMed on 31 August 2022, the Google Scholar and Cochrane Library on 3 September 2022, and the Embase on 15 October 2022. The identified studies were screened by two independent investigators. Studies assessing the diagnostic accuracy of AI-based devices for the detection of AF in adults against a gold standard were selected. Qualitative and quantitative synthesis to calculate the pooled sensitivity and specificity was performed, and the QUADAS-2 tool was used for the risk of bias and applicability assessment. We screened 14,770 studies, from which 31 were eligible and included. All were diagnostic accuracy studies with case-control or cohort design. The main technologies used were: (a) photoplethysmography (PPG) with pooled sensitivity 95.1% and specificity 96.2%, and (b) single-lead ECG with pooled sensitivity 92.3% and specificity 96.2%. In the PPG group, 0% to 43.2% of the tracings could not be classified using the AI algorithm as AF or not, and in the single-lead ECG group, this figure fluctuated between 0% and 38%. Our analysis showed that AI-based methods for the diagnosis of atrial fibrillation have high sensitivity and specificity for the detection of AF. Further studies should examine whether utilization of these methods could improve clinical outcomes.

Validation of Photoplethysmography Using a Mobile Phone Application for the Assessment of Heart Rate Variability in the Context of Heart Rate Variability-Biofeedback

OBJECTIVE

Heart rate variability-biofeedback (HRV-BF) is an effective intervention to reduce stress and anxiety and requires accurate measures of real-time HRV. HRV can be measured through photoplethysmography (PPG) using the camera of a mobile phone. No studies have directly compared HRV-BF supported through PPG against classical electrocardiogram (ECG). The current study aimed to validate PPG HRV measurements during HRV-BF against ECG.

METHODS

Fifty-seven healthy participants (70% women) with a mean (standard deviation) age of 26.70 (9.86) years received HRV-BF in the laboratory. Participants filled out questionnaires and performed five times a 5-minute diaphragmatic breathing exercise at different paces (range, ~6.5 to ~4.5 breaths/min). Four HRV indices obtained through PPG, using the Happitech software development kit, and ECG, using the validated NeXus apparatus, were calculated and compared: RMSSD, pNN50, LFpower, and HFpower. Resonance frequency (i.e., optimal breathing pace) was also compared between methods.

RESULTS

All intraclass correlation coefficient values of the five different breathing paces were "near perfect" (>0.90) for all HRV indices: lnRMSSD, lnpNN50, lnLFpower, and lnHFpower. All Bland-Altman analyses (with just three incidental exceptions) showed good interchangeability of PPG- and ECG-derived HRV indices. No systematic evidence for proportional bias was found for any of the HRV indices. In addition, correspondence in resonance frequency detection was good with 76.6% agreement between PPG and ECG.

CONCLUSIONS

PPG is a potentially reliable and valid method for the assessment of HRV. PPG is a promising replacement of ECG assessment to measure resonance frequency during HRV-BF.

Feasibility of using intermittent active monitoring of vital signs by smartphone users to predict SARS-CoV-2 PCR positivity

Early detection of highly infectious respiratory diseases, such as COVID-19, can help curb their transmission. Consequently, there is demand for easy-to-use population-based screening tools, such as mobile health applications. Here, we describe a proof-of-concept development of a machine learning classifier for the prediction of a symptomatic respiratory disease, such as COVID-19, using smartphone-collected vital sign measurements. The Fenland App study followed 2199 UK participants that provided measurements of blood oxygen saturation, body temperature, and resting heart rate. Total of 77 positive and 6339 negative SARS-CoV-2 PCR tests were recorded. An optimal classifier to identify these positive cases was selected using an automated hyperparameter optimisation. The optimised model achieved an ROC AUC of 0.695 ± 0.045. The data collection window for determining each participant's vital sign baseline was increased from 4 to 8 or 12 weeks with no significant difference in model performance (F(2) = 0.80, p = 0.472). We demonstrate that 4 weeks of intermittently collected vital sign measurements could be used to predict SARS-CoV-2 PCR positivity, with applicability to other diseases causing similar vital sign changes. This is the first example of an accessible, smartphone-based remote monitoring tool deployable in a public health setting to screen for potential infections.

Resting heart rate is a population-level biomarker of cardiorespiratory fitness: The Fenland Study

INTRODUCTION

Few large studies have evaluated the relationship between resting heart rate (RHR) and cardiorespiratory fitness. Here we examine cross-sectional and longitudinal relationships between RHR and fitness, explore factors that influence these relationships, and demonstrate the utility of RHR for remote population monitoring.

METHODS

In cross-sectional analyses (The UK Fenland Study: 5,722 women, 5,143 men, aged 29-65y), we measured RHR (beats per min, bpm) while seated, supine, and during sleep. Fitness was estimated as maximal oxygen consumption (ml⋅min-1⋅kg-1) from an exercise test. Associations between RHR and fitness were evaluated while adjusting for age, sex, adiposity, and physical activity. In longitudinal analyses (6,589 participant subsample), we re-assessed RHR and fitness after a median of 6 years and evaluated the association between within-person change in RHR and fitness. During the coronavirus disease-2019 pandemic, we used a smartphone application to remotely and serially measure RHR (1,914 participant subsample, August 2020 to April 2021) and examined differences in RHR dynamics by pre-pandemic fitness level.

RESULTS

Mean RHR while seated, supine, and during sleep was 67, 64, and 57 bpm. Age-adjusted associations (beta coefficients) between RHR and fitness were -0.26, -0.29, and -0.21 ml⋅kg-1⋅beat-1 in women and -0.27, -0.31, and -0.19 ml⋅kg-1⋅beat-1 in men. Adjustment for adiposity and physical activity attenuated the RHR-to-fitness relationship by 10% and 50%, respectively. Longitudinally, a 1-bpm increase in supine RHR was associated with a 0.23 ml⋅min-1⋅kg-1 decrease in fitness. During the pandemic, RHR increased in those with low pre-pandemic fitness but was stable in others.

CONCLUSIONS

RHR is a valid population-level biomarker of cardiorespiratory fitness. Physical activity and adiposity attenuate the relationship between RHR and fitness.

Engagement With mHealth COVID-19 Digital Biomarker Measurements in a Longitudinal Cohort Study: Mixed Methods Evaluation

BACKGROUND

The COVID-19 pandemic accelerated the interest in implementing mobile health (mHealth) in population-based health studies, but evidence is lacking on engagement and adherence in studies. We conducted a fully remote study for ≥6 months tracking COVID-19 digital biomarkers and symptoms using a smartphone app nested within an existing cohort of adults.

OBJECTIVE

We aimed to investigate participant characteristics associated with initial and sustained engagement in digital biomarker collection from a bespoke smartphone app and if engagement changed over time or because of COVID-19 factors and explore participants' reasons for consenting to the smartphone substudy and experiences related to initial and continued engagement.

METHODS

Participants in the Fenland COVID-19 study were invited to the app substudy from August 2020 to October 2020 until study closure (April 30, 2021). Participants were asked to complete digital biomarker modules (oxygen saturation, body temperature, and resting heart rate [RHR]) and possible COVID-19 symptoms in the app 3 times per week. Participants manually entered the measurements, except RHR that was measured using the smartphone camera. Engagement was categorized by median weekly frequency of completing the 3 digital biomarker modules (categories: 0, 1-2, and ≥3 times per week). Sociodemographic and health characteristics of those who did or did not consent to the substudy and by engagement category were explored. Semistructured interviews were conducted with 35 participants who were purposively sampled by sex, age, educational attainment, and engagement category, and data were analyzed thematically; 63% (22/35) of the participants consented to the app substudy, and 37% (13/35) of the participants did not consent.

RESULTS

A total of 62.61% (2524/4031) of Fenland COVID-19 study participants consented to the app substudy. Of those, 90.21% (2277/2524) completed the app onboarding process. Median time in the app substudy was 34.5 weeks (IQR 34-37) with no change in engagement from 0 to 3 months or 3 to 6 months. Completion rates (≥1 per week) across the study between digital biomarkers were similar (RHR: 56,517/77,664, 72.77%; temperature: 56,742/77,664, 73.06%; oxygen saturation: 57,088/77,664, 73.51%). Older age groups and lower managerial and intermediate occupations were associated with higher engagement, whereas working, being a current smoker, being overweight or obese, and high perceived stress were associated with lower engagement. Continued engagement was facilitated through routine and personal motivation, and poor engagement was caused by user error and app or equipment malfunctions preventing data input. From these results, we developed key recommendations to improve engagement in population-based mHealth studies.

CONCLUSIONS

This mixed methods study demonstrated both high initial and sustained engagement in a large mHealth COVID-19 study over a ≥6-month period. Being nested in a known cohort study enabled the identification of participant characteristics and factors associated with engagement to inform future applications in population-based health research.