FibriCheck Detection Capabilities for Atrial Fibrillation (FDA – AF): A Multicenter Validation Study

FibriCheck Detection Capabilities for Atrial Fibrillation (FDA – AF): A Multicenter Validation Study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article FibriCheck Detection Capabilities for Atrial Fibrillation (FDA – AF): A Multicenter Validation Study John Sollee, Baljash Cheema, David Slotwiner, Alexander Volodarskiy, and 9 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6849469/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 20 Nov, 2025 Read the published version in npj Digital Medicine → Version 1 posted 9 You are reading this latest preprint version Abstract Background Atrial fibrillation (AF) is the most common arrhythmia worldwide and is associated with significant morbidity, mortality, and healthcare spending. Despite medical advances, AF remains underdiagnosed and undertreated, leading to preventable complications. FibriCheck © [Qompium NV, Hasselt, Belgium] is a medical analysis platform that uses an end-to-end algorithm to detect AF based on photoplethysmography (PPG) signals recorded on consumer smartphones. Purpose The study aimed to validate FibriCheck in a large, multi-center and multi-national cohort on ten popular smartphone devices. Methods A total of 236 patients were recruited from five independent, large academic centers in the United States and Europe. The FibriCheck system incorporates several convolutional neural networks to detect individual heartbeats, estimate average heart rate, and classify the rhythm based on PPG signals. Classification is verified by a FibriCheck technician. Classification performance was compared to the standard 12-lead electrocardiogram in the study population. Performance was assessed across clinical subgroups and smartphone devices. Results FibriCheck demonstrated high overall accuracy and reliability in detecting AF without technician verification: accuracy 98.5% (95% CI: 98.0%-99.0%); sensitivity 96.3% (95% CI: 94.4%-97.7%); specificity 99.3% (95% CI: 98.8%-99.7%); positive predictive value 98.0% (95% CI: 96.5%-98.9%); negative predictive value 99.8% (95% CI: 99.6%-99.9%). Performance was not affected by smartphone device or the presence or absence of comorbid heart failure, vascular disease, hypertension, diabetes, or stroke. Sensitivity was reduced in those with darker skin tone and higher BMI, but this was mitigated by technician verification. Conclusions The study confirms the high accuracy, sensitivity, and specificity of the FibriCheck algorithm in detecting AF across various smartphone models and clinical subgroups. These findings support the use of FibriCheck as a reliable, low-cost, and easily accessible tool for AF detection in a diverse patient population. Health sciences/Diseases/Cardiovascular diseases/Arrhythmias/Atrial fibrillation Biological sciences/Computational biology and bioinformatics/Machine learning atrial fibrillation artificial intelligence consumer devices Figures Figure 1 INTRODUCTION Atrial fibrillation (AF) is the most common arrhythmia worldwide, which has significantly increased in both incidence and prevalence over the last 50 years, reaching the level of a cardiovascular disease (CVD) epidemic in the 21st century. 1 – 3 Rising AF burden is driven by population aging 4 and increased rates of multimorbidity such as obesity, diabetes, hypertension (HTN), and chronic stress. 3 AF is causally associated with increased risk of myocardial infarction (MI), embolic ischemic stroke, heart failure (HF), and chronic kidney disease (CKD). 5 , 6 Development and progression of AF and its associated comorbidities are interdependent. 7 , 8 For instance, a sub-analysis of the Framingham Heart Study demonstrated that 37% of patients with new AF had HF, and conversely, 57% of patients with new HF had preexisting AF. 8 The prevalence of one condition was associated with a higher incidence of the other. 8 Patients with AF also have a 5 times increased risk of stroke, the leading cause of chronic severe disability in the US and the 5th leading cause of death. 9 Despite medical advances, AF often remains underdiagnosed, leading to preventable complications and mortality. Because AF is typically diagnosed by in-office 12-lead electrocardiogram (ECG), paroxysmal variants and asymptomatic cases are often missed. 10 Up to 20% of patients presenting with AF-related strokes are undiagnosed, 7 over 90% of whom meet criteria for chronic oral anticoagulation. 11 In short, there remains an ongoing need to develop clinically and economically feasible methods for early and accurate AF detection and monitoring to improve global public health. With recent advancements in microchips, sensor technologies, and cloud computing, researchers have developed a wide variety of tools for remote healthcare monitoring. Most recent innovations in the field of CVD and beyond have focused on consumer wearables or mobile smartphone applications, many of which incorporate artificial intelligence (AI). 12 , 13 Photoplethysmography (PPG) has emerged as the preferred signal modality for measuring heart rate and detecting arrhythmias, as smartphones and wearables already incorporate capable sensors. 14 – 16 In PPG, diodes emit light towards human tissue, and the reflected light is captured by photosensors. Because the intensity and pulsatility of the reflected light is a function of the propagation of arterial pressure pulses within the microvascular bed, PPG signals provide valuable real-time information about cardiovascular health, including oxygen saturation, heart rate, blood pressure, and cardiac output. 17 PPG-based AF detection algorithms typically work by extracting temporal, morphological, and/or spectral features from raw PPG signals, which are subsequently input into a classifier. 15 A recent review article identified 24 studies incorporating PPG for AF detection, half of which incorporated either machine or deep learning techniques. 15 FibriCheck © [Qompium NV, Hasselt, Belgium] is a CE Class IIa and FDA-cleared medical analysis platform that uses an end-to-end algorithm to classify heart rhythms based on PPG signals collected using their smartphone application. Prior studies, many from the multicenter European TeleCheck-AF project, demonstrated high usability, compliance, and patient satisfaction ratings, 18 , 19 including in the primary care setting. 20 Small, and/or single-center studies (N ≤ 300) have demonstrated excellent performance in AF detection, 21 – 24 including in real-world conditions, 25 with the most recent validation study showing 100% sensitivity, 98.9% specificity, and 99.2% accuracy across 122 participants at a single European center. 26 Larger, multicenter validation studies in more diverse populations including non-European participants have not been performed. The FDA-AF study aimed to validate FibriCheck in a large, multi-center and multi-national cohort on the ten most popular smartphone devices. METHODS Study Design and Data Acquisition The study was performed across five independent, large academic medical centers in the United States (US) and Europe: University Hospital Antwerp (UZA), Belgium; Hospital Oost-Limburg Genk (ZOL), Belgium; University of Oklahoma Health Sciences Center, Oklahoma, US; Northwestern Medicine, Chicago, US; New York Presbyterian Queens, New York, US. The institutional review board of each institution independently approved the study, and the study followed all principles of the Declaration of Helsinki (7th edition, October 2013), per the International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use – Good Clinical Practice (ICH-GCP) guidelines. Written informed consent was obtained for all participants. Every attempt was made to protect patient confidentiality, and participants had the right to withdraw from the study at any time. Participants were eligible for inclusion if they met the following criteria: at least 22 years of age, capable of independently performing FibriCheck readings under observation from the study team and receiving active cardiology care either in the outpatient setting or hospitalized with or without AF. Enrollment was expected to last one month and attain 50% AF prevalence with 15–20% class IV or higher on the Fitzpatrick scale to adequately sample patients with darker skin tone (a known population where PPG signal interpretation can be less accurate). 27 Participants were excluded if they had implantable pacemakers, cardioverter-defibrillators, or other electric devices, as such devices could interfere with natural heart rhythm; the model was not trained on such cases, and the FibriCheck platform is not intended for use in such patients. Participants were also excluded if they were unable to complete measurements independently due to physical or medical constraints, actively enrolled in other clinical trials, and/or pregnant or nursing women. Data acquisition for consented patients was performed at outpatient cardiology clinics or in hospital if patients were admitted. For each participant, the following demographic and clinical data were recorded: US or non-US citizen, BMI ≥ 30 or < 30, skin type (based on the Fitzpatrick scale), presence or absence of a history of AF, HF, vascular disease, hypertension (HTN), diabetes, and stroke. The FibriCheck application with access to the FibriCheck cloud and FibriCheck algorithm (v1.5.2) was pre-installed on 10 different iOS (n = 8) and Android (n = 2) devices. Participants were instructed to sequentially place their fingers on the camera of each of the 10 devices to allow for 60-second PPG reading as outlined in Fig. 1 . The number of recording attempts per smartphone device was determined by the FibriCheck quality analysis, described below. If a PPG recording was deemed to have insufficient quality according to the FibriCheck algorithm, participants were instructed to repeat the recording until adequate quality was achieved, with a maximum of three attempts permitted per smartphone. At the end of each recording, the PPG data was automatically transmitted to the FibriCheck Cloud for processing and analysis by the FibriCheck algorithm. A 12-lead electrocardiogram (ECG) was used as the reference standard. ECGs were performed for each participant twice during PPG recordings, once during the third smartphone PPG recording (Apple iPhone 15) and once during the eighth smartphone PPG recording (Samsung Galaxy A53) (Fig. 1 ). Each ECG was evaluated by at least two board certified and independent cardiac electrophysiologists and labeled as regular, AF, atrial flutter, or unclassified (not one of the three other rhythms). If there were discrepancies in the findings of the two experts, then a third expert was consulted, and the majority decision decided the reference diagnosis. In the case of disagreement among all three experts or if the ECG was deemed unreadable (poor quality), the data were excluded from the analysis. If the two ECGs for a single patient were deemed to be different rhythms, then the data were also excluded. FibriCheck Platform Algorithm The FibriCheck platform uses an end-to-end algorithm incorporating three convolutional neural networks (CNN): (1) quality detection , (2) heartbeat detection , and (3) rhythm classification . To initiate data collection, users place their finger on the smartphone camera lens. Once the presence of a finger is confirmed using a dedicated detection algorithm, a video is recorded in YUV color format for one minute. After recording, the video is converted to the RGB color format, where the RGB components are treated as potential PPG signals. The RGB signals then undergo a series of preprocessing steps, which include noise filtering, derivative calculation, normalization, and signal truncation. These preprocessing steps are aimed at enhancing the quality of the extracted PPG signals and reducing the influence of noise or artifacts. Following preprocessing, the (1) quality detection CNN indicates if specific segments within the PPG are too noisy for further analysis. If the model determines that more than 30 seconds of the PPG signal is too noisy or fails to meet quality standards, the measurement is flagged as "insufficient quality," and no further clinical analysis is performed. Sufficient quality signals are passed to the (2) heartbeat detection CNN, which indicates the location of heartbeats in the preprocessed PPG signal and constructs a PPG-based tachogram and average heart rate measurement over the one-minute measurement. Both the (1) quality detection and (2) heartbeat detection models are trained on beat-to-beat annotated internal PPG datasets consisting of tens of thousands of synchronized PPG-ECG data samples. When the signal quality meets the required criteria and the heart rate falls within the validated range, the platform proceeds to the (3) rhythm classification CNN. This model has been trained on a diverse dataset of over one million rhythm-annotated measurements, encompassing various heart rhythm disorders, including AF and atrial flutter amongst others. The algorithm then classifies the heart rhythm as regular, possible AF, or unclassified rhythm. Technician Verification In addition to fully automated classification by the rhythm classification CNN as above, each PPG recording of sufficient quality is independently reviewed by a blinded FibriCheck technician. Verification takes place within 48 hours and happens on a per-measurement basis: each measurement is first analyzed by the algorithm and then automatically queued for human verification. By visually reviewing the PPG recordings, the technician classifies the rhythm as “regular,” “possible AF,” or “unclassified rhythm” based on criteria extrapolated from the peer-reviewed practical guidance on signal interpretation and clinical scenarios from TeleCheck-AF. 28 Criteria for a regular rhythm include equal intervals between peaks in the raw PPG signal, sporadic irregularity, a single line or wave-like pattern in the tachogram, and a dense or ellipse-shaped cluster in the Poincaré plot. Criteria for possible AF are irregularly varying intervals between the peaks in the PPG tracing, randomly distributed points on the tachogram, and the absence of a distinct cluster of points on the Poincaré plot. If the recording does not meet criteria for either regular rhythm or possible AF, then it is labeled an unclassified rhythm. In the case of discrepancies between the automated and human classification, the automated classification is overruled, and the final classification is adjusted accordingly. Statistical Analysis and Outcome Measures Sample size calculations were performed prior to subject recruitment. Based on European post-market surveillance data, an accuracy of 95.8% (95% CI: 93.76–97.32%) can be expected in detecting AF. For the analysis, the reference value (p0) was set to the lower bound of the accuracy CI, which was 93.8%. The expected accuracy of the FibriCheck system was set at p = 0.990 based on European data. Using the exact Clopper & Pearson method and aiming for a significance level (alpha) of 0.025 and a power (1-beta) of 0.8, the calculated sample size required for this comparison was determined to be 114 recordings of sufficient signal quality. To allow for potential sub analyses, the targeted number of participants was set at 250. The results obtained from the automated analysis by the FibriCheck system were compared with the 12-lead ECG reference diagnosis to evaluate the ability of the system to correctly differentiate the heart rhythm. Continuous variables were presented as mean with standard deviation (SD) or median with interquartile range (IQR). Categorical values were reported as counts and percentages. The performance of FibriCheck was evaluated by calculating the accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) along with corresponding 95% CI. Prevalence-adjusted PPV and NPV were also calculated assuming an AF prevalence of 6%. Data was analyzed using MedCalc Software Ltd.’s Diagnostic Test Evaluation Calculator (Version 22.016). The main outcome was overall performance across all devices and participants. Sub-group analyses were conducted to determine performance in individual devices, across different skin tones, in those with BMI over and under 30, and in those with and without a previous AF, HF, HTN, diabetes, vascular disease, or stroke diagnosis. Comparative Analysis with FDA-Cleared Devices To benchmark FibriCheck's performance, a comparative analysis was conducted using the FDA's 510(k) premarket notification database (Table 6 ). The analysis included 22 devices with similar indications for use, focusing on self-testing by patients diagnosed with or susceptible to AF. Among these, seven devices reported clinical testing results. RESULTS Study Population A total of 252 participants were enrolled. Of these, 16 (6.4%) were excluded from the study, including 4 who met exclusion criteria but were inappropriately enrolled, 5 dropouts due to withdrawn informed consent, 3 due to unavailability of a 12-lead ECG device during data acquisition, 2 due to interruption of the study by other medical examinations, 1 due to poor quality ECG, and 1 due to different rhythms being detected on the two ECG recordings for that single patient. Therefore, a total of 236 participants were included in the analysis. Demographics and clinical information are shown in Table 1 . Among the 236 participants, 157 (66.5%) were identified as having regular rhythm, and 60 (25.4%) were identified as having AF based on 12-lead ECG (reference diagnosis). Atrial flutter was identified in 10 (4.2%) participants, and the remaining 9 (3.8%) participants presented with unclassified rhythms. For the primary analysis, atrial flutter recordings were excluded (99 recordings). Additionally, 48 recordings were flagged by the quality assessment algorithm and excluded prior to heartbeat detection and rhythm classification. Therefore, a total of 2,195 recordings were available for the primary analysis. Median time from PPG recording to capture of ECG reference diagnosis was 2 minutes (IQR 2,3). Table 1 Participant Demographics and Clinical Characteristics. Total population (n = 236) US population (n = 158) Non-US population (n = 78) Age (in years) Median (Q1-Q3) 65 (54–74) 65 (50–75) 66 (58.8–72) Sex, n (%) Male 143 (60.6%) 86 (54.4%) 57 (73.1%) Female 93 (39.4%) 72 (45.6%) 21 (26.9%) Body Mass Index (kg/m 2 ) Median (Q1-Q3) 28 (25–32) 29 (25–33) 27 (24.8–30) Skin tone, Fitzpatrick Scale Type I, n (%) 47 (19.9%) 21 (13.3%) 26 (33.3%) Type II, n (%) 113 (47.9%) 71 (44.9%) 42 (53.8%) Type III, n (%) 28 (11.9%) 20 (12.7%) 8 (10.3%) Type IV, n (%) 16 (6.8%) 15 (9.5%) 1 (1.3%) Type V, n (%) 15 (6.4%) 14 (8.9%) 1 (1.3%) Type VI, n (%) 17 (7.2%) 17 (10.8%) 0 (0.0%) Medical history Atrial fibrillation, n (%) 148 (62.7%) 86 (54.4%) 62 (79.5%) Persistent atrial fibrillation 68 (28.8%) 31 (19.6%) 25 (32.1%) Paroxysmal atrial fibrillation 80 (33.9%) 55 (34.8%) 37 (47.4%) Heart failure, n (%) 67 (28.4%) 59 (37.3%) 8 (10.3%) Vascular disease, n (%) 27 (11.4%) 22 (13.9% 5 (6.4%) Hypertension, n (%) 125 (53.0%) 99 (62.7%) 26 (33.3%) Diabetes, n (%) 53 (22.5%) 42 (26.6%) 11 (14.1%) Stroke, n (%) 36 (15.3%) 27 (17.1%) 9 (11.5%) COPD diagnosis, n (%) 15 (6.4%) 10 (6.3%) 5 (6.4%) Primary Analysis: Overall Performance and Performance by Smartphone Device The FibriCheck algorithm demonstrated high overall accuracy and reliability in differentiating AF from non-AF (Tables 2 and 3 ). Without technician verification, accuracy was 98.5% (95% CI: 98.0%-99.0%), sensitivity was 96.3% (95% CI: 94.4%-97.7%), specificity was 99.3% (95% CI: 98.8%-99.7%), PPV was 98.0% (95% CI: 96.5%-98.9%), and NPV was 99.8% (95% CI: 99.6%-99.9%). Performance was not significantly changed with technician verification (Table 2 ). The FibriCheck algorithm demonstrated high performance in all 10 smartphone devices, with highest accuracy in the iPhone 13 Pro (100%) but no significant differences among all devices. Detailed performance by smartphone device is shown in Table 3 . Table 2 FibriCheck Overall Performance Metrics. Performance Metric FibriCheck without Verification FibriCheck with Verification Accuracy (95% CI) 98.5% (98.0%-99.0%) 98.8% (98.2%-99.2%) Sensitivity (95% CI) 96.3% (94.4%-97.7%) 99.0% (97.7%-99.6%) Specificity (95% CI) 99.3% (98.8%-99.7% 98.7% (98.0%-99.2%) Disease prevalence 26.1% 26.2% PPV (95% CI) 98.0% (96.5%-98.9%) 96.4% (94.7%-97.6%) Prevalence-adjusted PPV, 6% 90.1% (83.4%-94.2%) 82.9% (76.1%-88.2%) NPV (95% CI) 98.7% (98.1%-99.2%) 99.6% (99.2%-99.8%) Prevalence-adjusted NPV, 6% 99.8% (99.6%-99.9%) 99.9% (99.9%-100%) Table 3 FibriCheck Performance by Smartphone Device Smartphone Device Accuracy Possible AF vs non-possible AF Overall performance (95% CI) Eligible for rhythm analysis*, n (%) 98.5% (98.0%-99.0%) 2,195/2,243 (98.2%) iPhone SE (3rd gen) (95% CI) ΔiPhone SE (3rd gen) - overall performance Eligible for rhythm analysis*, n (%) Time until reference diagnosis in minutes, median (Q1-Q3) 99.1% (96.8%-99.9%) + 0.6% 224/226 (99.1%) 6 (4,9) iPhone 15 Pro (95% CI) ΔiPhone 15 Pro - overall performance Eligible for rhythm analysis*, n (%) Time until reference diagnosis in minutes, median (Q1-Q3) 98.2% (95.4%-99.5%) -0.3% 218/225 (96.9%) 5 (2,7) iPhone 15 (95% CI) ΔiPhone 15 - overall performance Eligible for rhythm analysis*, n (%) Time until reference diagnosis in minutes, median (Q1-Q3) 96.8% (93.6%-98.7%) -1.7% 221/226 (97.8%) N/A, simultaneous iPhone 11 (95% CI) ΔiPhone 11 - overall performance Eligible for rhythm analysis*, n (%) Time until reference diagnosis in minutes, median (Q1-Q3) 98.2% (95.5%-99.5%) -0.3% 224/226 (99.1%) 3 (3,4) iPhone 14 (95% CI) ΔiPhone 14 - overall performance Eligible for rhythm analysis*, n (%) Time until reference diagnosis in minutes, median (Q1-Q3) 99.6% (97.5%-100%) + 1.1% 221/225 (98.2%) 4 (3,5) iPhone 12 (95% CI) ΔiPhone 12 - overall performance Eligible for rhythm analysis*, n (%) Time until reference diagnosis in minutes, median (Q1-Q3) 98.7% (96.1%-99.7%) + 0.2% 223/225 (99.1%) 2 (1,3) Samsung Galaxy S23 (95% CI) ΔSamsung Galaxy S23 - overall performance Eligible for rhythm analysis*, n (%) Time until reference diagnosis in minutes, median (Q1-Q3) 97.2% (94.0%-99.0%) -1.3% 213/221(96.4%) 2 (2,3) Samsung Galaxy A54 (95% CI) ΔSamsung Galaxy A54 - overall performance Eligible for rhythm analysis*, n (%) Time until reference diagnosis in minutes, median (Q1-Q3) 98.2% (95.4%-99.5%) -0.3% 218/223 (97.8%) N/A, simultaneous iPhone 13 (95% CI) ΔiPhone 13 - overall performance Eligible for rhythm analysis*, n (%) Time until reference diagnosis in minutes, median (Q1-Q3) 99.5% (97.5%-100%) + 1.0% 217/222 (97.7%) 2 (2,3) iPhone 13 Pro (95% CI) ΔiPhone 13 Pro - overall performance Eligible for rhythm analysis*, n (%) Time until reference diagnosis in minutes, median (Q1-Q3) 100% (98.3%-100%) + 1.5% 216/224 (96.4%) 3 (3,4) Subgroup Analyses Summary of performance in each subgroup is shown in Tables 4 and 5 . The number of recordings that were excluded due to atrial flutter or poor quality is also shown in the table. The FibriCheck algorithm demonstrated high accuracy across all skin tones, although sensitivity was lower in participants with darker skin tones (Fitzpatrick type IV or higher). Verification by a FibriCheck technician successfully mitigated the risks related to the lower sensitivity of 79.6% (65.7% − 89.8%) in participants with a dark skin tone compared to participants with a pale or medium skin tone. The FibriCheck algorithm also showed high accuracy and reliability in participants with a previous AF diagnosis. The system performed consistently well in participants with and without HF, vascular disease, HTN, diabetes, and stroke. Table 4 Sub-group Analysis Performance without Technician Verification. Possible AF vs non-possible AF Sub-group BMI ≥ 30 BMI < 30 Pale skin Medium skin Dark skin Diabetes No diabetes AF No AF HF No HF Stroke No Stroke HTN No HTN Vascular disease No vascular disease Accuracy (95% CI) 97.2% (95.9%-98.2%) 99.5% (98.9%-99.8%) 98.7% (98.0%-99.2%) 99.5% (98.3%-99.9%) 96.3% (93.4%-98.1%) 97.8% (96.0%-98.9%) 98.8% (98.1%-99.2%) 97.9% (96.9%-98.6%) - 97.5% (95.9%-98.6%) 98.9% (98.3%-99.4%) 97.6% (95.4%-99.0%) 98.7% (98.1%-99.2%) 98.1% (97.2%-98.8%) 99.0% (98.2%-99.5%) 100% (98.5%-100%) 98.4% (97.7%-98.9%) Sensitivity (95% CI) 93.7% (89.9%-96.5%) 98.2% (96.1%-99.3%) 97.6% (95.7%-98.8%) 100% (94.7%-100%) 79.6% (65.7%-89.8%) 93.6% (88.8%-96.8%) 97.5% (95.5%-98.8%) 96.3% (94.4%-97.7%) - 96.0% (92.3%-98.3%) 96.5% (94.1%-98.1%) 89.9% (81.0%-95.5%) 97.4% (95.5%-98.6%) 96.1% (93.6%-97.8%) 96.8% (93.2%-98.8%) ​​100% (95.1%-100%) 95.8% (93.6%-97.4%) Specificity (95% CI) 98.5% (97.2%-99.3%) 99.90% (99.4%-100%) 99.2% (98.5%-99.7%) 99.4% (98.0%-99.9%) 99.6% (97.8%-100%) 100% (98.9%-100%) 99.2% (98.5%-99.6%) 99.0% (98.0%-99.6%) 99.6% (98.7%-99.9%) 98.3% (96.5%-99.3%) 99.7% (99.2%-99.9%) 100% (98.6%-100%) 99.2% (98.6%-99.6%) 99.1% (98.2%-99.6%) 99.5% (98.8%-99.9%) 100% (97.8%-100%) 99.3% (98.7%-99.6%) Excluded (atrial flutter, poor quality) 44 (20, 24) 103 (79, 24) 106 (69, 37) 106 (69, 37) 106 (69, 37) 33 (20, 13) 114 (79, 35) 117 (79, 38) 30 (20, 10) 51 (40, 11) 96 (59, 37) 132 (89, 43) 132 (89, 43) 67 (40, 27) 80 (59, 21) 24 (20, 4) 123 (79, 44) Table 5 Sub-group Analysis Performance with Technician Verification. Possible AF vs non-possible AF Sub-group BMI ≥ 30 BMI < 30 Pale skin Medium skin Dark skin Diabetes No diabetes AF No AF HF No HF Stroke No Stroke HTN No HTN Vascular disease No vascular disease Accuracy (95% CI) 98.2% (97.1%-99.0%) 99.2% (98.5%-99.6%) 98.7% (98.0%-99.2%) 99.0% (97.6%-99.7%) 98.6% (96.6%-99.6%) 99.0% (97.6%-99.7%) 98.7% (98.1%-99.2%) 98.5% (97.7%-99.1%) - 97.9% (96.4%-98.9%) 99.1% (98.5%-99.5%) 99.4% (97.9%-99.9%) 98.7% (98.0%-99.1%) 99.0% (98.2%-99.5%) 98.5% (97.6%-99.2%) 99.2% (97.0%-99.9%) 98.7% (98.1%-99.2%) Sensitivity (95% CI) 98.8% (96.4%-99.7%) 99.1% (97.4%-99.8%) 99.3% (98.1%-99.9%) 100% (94.7%-100%) 93.8% (82.8%-98.7%) 98.8% (95.8%-99.9%) 99.0% (97.5%-99.7%) 99.0% (97.7%-99.6%) - 98.0% (95.0%-99.5%) 99.5% (98.1%-99.9%) 97.4% (91.0%-99.7%) 99.2% (98.0%-99.8%) 99.2% (97.7%-99.8%) 98.4% (95.5%-99.7%) 100% (95.1%-100%) 98.8% (97.4%-99.6%) Specificity (95% CI) 98.0% (96.6%-98.9%) 99.2% (98.4%-99.7%) 98.4% (97.5%-99.1%) 98.9% (97.1%-99.7%) 99.6% (97.8%-100%) 99.1% (97.3%-99.8%) 98.6% (97.8%-99.2%) 98.2% (97.0%-99.0%) 99.2% (98.3%-99.7%) 97.8% (95.8%-99.0%) 99.0% (98.3%- 99.5%) 100% (98.6%-100%) 98.5% (97.6%-99.0%) 98.9% (97.8%-99.5%) 98.6% (97.5%-99.3%) 98.8% (95.7%-99.9%) 98.7% (98.0%-99.2%) Excluded (atrial flutter, poor quality) 44 (20, 24) 103 (79, 24) 106 (69, 37) 106 (69, 37) 106 (69, 37) 33 (20, 13) 114 (79, 35) 117 (79, 38) 30 (20, 10) 51 (40, 11) 96 (59, 37) 132 (89, 43) 132 (89, 43) 67 (40, 27) 80 (59, 21) 24 (20, 4) 123 (79, 44) Comparative Analysis with FDA-Cleared Devices Compared to the 22 devices with similar indications for use, FibriCheck demonstrated superior or equivocal sensitivity and specificity. Full resulted are shown in Table 6 . Table 6 Overview of reported performance metrics of previously cleared devices with a similar indication for use and reported clinical performance based on the 510(k) premarket notification database. Device name 510(k) number Sensitivity (95% CI) Specificity (95% CI) FibriCheck K232804 96.3% (94.4%-97.7%) 99.3% (98.8%-99.7%) FibriCheck K173872 95.60% (no 95% CI reported) 96.6% (no 95% CI reported) Coala Heart Monitor K182040 97.2% (no 95% CI reported) 94.6% (no 95% CI reported) Study Watch with Irregular Pulse Monitor K192415 85% (79%-90%) 96% (93%-99%) Halo AF Detection System K201208 93.3% (no 95% CI reported) 99.1% (no 95% CI reported) Scan Monitor K201456 96.3% Lower bound 95% CI: 89.4%, No upper bound reported 100% Lower bound 95% CI: 96.7% No upper bound reported Study Watch with Irregular Pulse Monitor (Home) Study Watch with Irregular Pulse Monitor K213357 96.1% (92.7%-98.0%) 98.1% (97.2%-99.1%) Withings Scan Monitor 2.0 K230812 99% (93%-100%) 99% (97%-100%) DISCUSSION The FDA-AF study validates the FibriCheck platform as a highly accurate and reliable tool for detecting AF in a diverse patient population. By demonstrating consistent performance across 10 of the most common smartphone devices, the study also underscores the platform’s ease of implementation and potential as a resource-efficient method for AF detection and monitoring outside of the clinical setting. The FibriCheck platform offers several clinical advantages over existing methods for AF detection and monitoring. Unlike traditional 12-lead ECGs, FibriCheck measurements can be performed at any time, within 60 seconds, and utilizing a device already owned by most patients. Patients without a formal diagnosis of AF but exhibiting symptoms may be instructed by a clinician to initiate FibriCheck readings when feeling symptomatic. Similarly, those with paroxysmal AF may be instructed to take periodic readings to assess AF burden. For select paroxysmal patients who are managed with a “pill in the pocket” approach, FibriCheck could guide self-administration of single dose antiarrhythmics (e.g., flecainide or propafenone) to promptly terminate the arrythmia. 29 FibriCheck can also be used to monitor for recurrence following electrical cardioversion or ablation procedures. 30 , 31 Like other patient-activated wearables including smartwatches and handheld ECG devices, FibriCheck may miss transient or asymptomatic arrythmias. 13 Still, unlike continuous monitoring devices such as the ZioPatch, 32 Holter monitor, or loop recorder, FibriCheck is entirely non-invasive, does not require external battery packs or chest leads, and can record and transmit unlimited readings without the need for repeat office visits or hardware exchanges. 13 Wrist-worn devices for continuous AF monitoring are being developed, such as the recently FDA-cleared Verily Study Watch; 33 however, these require the purchase of additional hardware rather than operating through a basic smartphone. In contrast, FibriCheck operates on devices already widely available, making it particularly suitable for resource-limited settings The FibriCheck algorithm performed well across a diverse patient population, reinforcing clinical utility, particularly given high rates of multimorbidity in AF. 34 , 35 Several studies have demonstrated that certain comorbidities and patient characteristics, most notably obesity and skin tone, can significantly affect PPG signal quality and lead to inaccurate biophysical measurements. Skin tone is often described using the Fitzpatrick scale, which classifies skin types from I, the lightest, to VI, the darkest based on response to ultraviolet light. 27 Monte Carlo simulations have shown that the AC/DC ratio of PPG signals, a measure of blood volume pulsatility detection, is compromised in darker skin (higher Fitzpatrick scale) due to increased light absorption by melanin. 36 , 37 This effect has been shown to result in signal loss in existing commercial wearables, including the Apple Watch series 5 and Fitbit Versa 2. 37 Obesity also affects PPG signal quality due to the effects of adipose and dermal tissue on penetration and scattering of light, 38 with effects on AC/DC signal degradation up to 40%. 39 Vascular disease and HTN have also been shown to affect PPG signals, however likely to a lesser extent. 40 The subgroup analysis demonstrated that FibriCheck maintains high accuracy, sensitivity, and specificity in patients with diabetes and prior stroke as well as pre-existing diagnoses of HF, HTN, and vascular disease. Sensitivity was reduced in those with darker skin tone, but this was mitigated by FibriCheck technician verification; with verification, sensitivity improved from 79.6–93.8%. Likewise, sensitivity was slightly reduced in those with BMI ≥ 30 kg/m 2 . This was also mitigated with technician verification, improving sensitivity from 93.7–98.8%. By offering technician verification, the FibriCheck platform can successfully mitigate the known effects of skin type and obesity on classification performance. This feature affords FibriCheck an advantage in comparison to other consumer platforms for mobile AF detection that do not offer human verification. To benchmark FibriCheck to the state-of-the-art, we performed a comparative analysis based on performance metrics of previously cleared devices with a similar indication for use and reported clinical performance based on the 510(k) premarket notification database. FibriCheck demonstrated comparable or superior performance to all identified devices reporting performance metrics. 33 , 41 In conclusion, the FDA-AF study confirms the high accuracy, sensitivity, and specificity of the FibriCheck algorithm in detecting AF across various smartphone platforms and clinical subgroups. These findings support the use of FibriCheck as a reliable, low-cost, and easily accessible tool for AF detection in a diverse patient population. Abbreviations AF: atrial fibrillation; PPG: photoplethysmography; CI: confidence interval; HTN: hypertension; HF: heart failure; CKD: chronic kidney disease; AI: artificial intelligence; ECG: electrocardiogram; CNN: convolutional neural network. Declarations Conflict of Interest Statement: JT: Consulting with GE Healthcare, Caption Health, Abbott, Eko Health. BJ: Consulting fees from Caption Health, Inc. and Viz.ai; served on an advisory board for Novo Nordisk, is an advisor with equity in Healthspan, Inc. and Zoe Biosciences; received speaking fees and honoraria from Bristol Meyers Squibb. All others report no relevant conflicts of interest. Author Contribution Writing - Original Draft and Visualization: J.S. Writing - Review & Editing: B.C., J.T. Investigation, Data Curation, and Project Administration: B.C., J.T., D. S., A.V., L.D., C.B., H.H., S.S., L.P., D.N., M.R-A., H.V.H. Supervision: J.T. Acknowledgments: The authors would like to thank the FibriCheck team at Qompium NV, Hasselt, Belgium. References Schnabel, R. B. et al. Fifty-Year Trends in Atrial Fibrillation Prevalence, Incidence, Risk Factors, and Mortality in the Community Renate. Lancet 386, 154–162 (2015). Chugh, S. S. et al. Worldwide epidemiology of atrial fibrillation: A global burden of disease 2010 study. Circulation 129, 837–847 (2014). Kornej, J., Börschel, C. S., Benjamin, E. J. & Schnabel, R. B. Epidemiology of Atrial Fibrillation in the 21st Century. Circ. Res. 127, 4–20 (2020). Heidenreich, P. A. et al. Forecasting the future of cardiovascular disease in the United States: A policy statement from the American Heart Association. Circulation 123, 933–944 (2011). Bisbal, F., Baranchuk, A., Braunwald, E., Bayés de Luna, A. & Bayés-Genís, A. Atrial Failure as a Clinical Entity: JACC Review Topic of the Week. J. Am. Coll. Cardiol. 75, 222–232 (2020). Benjamin, E. J. et al. Heart Disease and Stroke Statistics-2019 Update: A Report From the American Heart Association . Circulation vol. 139 (2019). Borowsky, L. H. et al. First Diagnosis of Atrial Fibrillation at the Time of Stroke. Cerebrovasc. Dis. 43, 192–199 (2017). Santhanakrishnan, R. et al. Atrial fibrillation begets heart failure and vice versa: Temporal associations and differences in preserved versus reduced ejection fraction. Circulation (2016) doi: 10.1161/CIRCULATIONAHA.115.018614 . January, C. T. et al. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: A report of the American College of cardiology/American heart association task force on practice guidelines and the heart rhythm society . Circulation vol. 130 (2014). Savelieva, I. & Camm, A. J. Clinical relevance of silent atrial fibrillation: Prevalence, prognosis, quality of life, and management. Journal of Interventional Cardiac Electrophysiology (2000) doi: 10.1023/A:1009823001707 . Turakhia, M. P. et al. Contemporary prevalence estimates of undiagnosed and diagnosed atrial fibrillation in the United States. Clin. Cardiol. 46, 484–493 (2023). Petek, B. J. et al. Consumer Wearable Health and Fitness Technology in Cardiovascular Medicine: JACC State-of-the-Art Review. J. Am. Coll. Cardiol. 82, 245–264 (2023). Spatz, E. S., Ginsburg, G. S., Rumsfeld, J. S. & Turakhia, M. P. Wearable Digital Health Technologies for Monitoring in Cardiovascular Medicine. N. Engl. J. Med. (2024) doi: 10.1056/nejmra2301903 . Tang, S. C. et al. Identification of Atrial Fibrillation by Quantitative Analyses of Fingertip Photoplethysmogram. Sci. Rep. 7, 1–7 (2017). Pereira, T. et al. Photoplethysmography based atrial fibrillation detection: a review. npj Digit. Med. 3, (2020). O’Sullivan, J. W. et al. Accuracy of Smartphone Camera Applications for Detecting Atrial Fibrillation A Systematic Review and Meta-analysis. JAMA Netw. Open 3, E202064 (2020). Almarshad, M. A., Islam, M. S., Al-Ahmadi, S. & Bahammam, A. S. Diagnostic Features and Potential Applications of PPG Signal in Healthcare: A Systematic Review. Healthc. 10, 1–28 (2022). Gawałko, M. et al. The European TeleCheck-AF project on remote app-based management of atrial fibrillation during the COVID-19 pandemic: Centre and patient experiences. Europace 23, 1003–1015 (2021). Gawałko, M. et al. Patient motivation and adherence to an on-demand app-based heart rate and rhythm monitoring for atrial fibrillation management: data from the TeleCheck-AF project. Eur. J. Cardiovasc. Nurs. 22, 412–424 (2023). Beerten, S. G., Proesmans, T. & Vaes, B. A heart rate monitoring app (fibricheck) for atrial fibrillation in general practice: Pilot usability study. JMIR Form. Res. 5, 1–9 (2021). Gruwez, H. et al. Head-to-head comparison of proprietary PPG and single-lead ECG algorithms for atrial fibrillation detection. EP Eur. 23, 2021 (2021). Selder, J. L. et al. Assessment of a standalone photoplethysmography (PPG) algorithm for detection of atrial fibrillation on wristband-derived data. Comput. Methods Programs Biomed . 197, (2020). Proesmans, T. et al. Mobile phone–based use of the photoplethysmography technique to detect atrial fibrillation in primary care: Diagnostic accuracy study of the fibricheck app. JMIR mHealth uHealth (2019) doi: 10.2196/12284 . Gruwez, H. et al. Evaluation of the device independent nature of a photoplethysmography-deriving smartphone app. EP Eur. 23, 2021 (2021). Gruwez, H. et al. Real-world validation of smartphone-based photoplethysmography for rate and rhythm monitoring in atrial fibrillation. Europace 26, 1–9 (2024). Wouters, F. et al. Comparative Evaluation of Consumer Wearable Devices for Atrial Fibrillation Detection: Validation Study. JMIR Form. Res. 9, 1–11 (2025). Gupta, V. & Sharma, V. K. Skin typing: Fitzpatrick grading and others. Clin. Dermatol. (2019) doi: 10.1016/j.clindermatol.2019.07.010 . Van Der Velden, R. M. J. et al. The photoplethysmography dictionary: Practical guidance on signal interpretation and clinical scenarios from TeleCheck-AF. Eur. Hear. J. - Digit. Heal. 2, 363–373 (2021). Joglar, J. A. et al. 2023 ACC/AHA/ACCP/HRS Guideline for the Diagnosis and Management of Atrial Fibrillation: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines . Circulation vol. 149 (2024). Calvert, P. et al. Remote rhythm monitoring using a photoplethysmography smartphone application after cardioversion for atrial fibrillation. Eur. Hear. J. - Digit. Heal. 5, 461–468 (2024). Gruwez, H. et al. Effectiveness of photoplethysmography-based rhythm monitoring after atrial fibrillation ablation using a smartphone application: DIGITOTAL study. Hear. Rhythm (2024) doi: 10.1016/j.hrthm.2024.11.026 . Turakhia, M. P. et al. Diagnostic utility of a novel leadless arrhythmia monitoring device. Am. J. Cardiol. (2013) doi: 10.1016/j.amjcard.2013.04.017 . Poh, M. Z. et al. Validation of a Deep Learning Algorithm for Continuous, Real-Time Detection of Atrial Fibrillation Using a Wrist-Worn Device in an Ambulatory Environment. J. Am. Heart Assoc. 12, 1–13 (2023). Arnett, D. K. et al. AHA/ACC/HHS strategies to enhance application of clinical practice guidelines in patients with cardiovascular disease and comorbid conditions from the American heart association, American college of cardiology, and US department of health and human servic. Circulation 130, 1662–1667 (2014). Alexander, K. P. et al. Outcomes of apixaban versus warfarin in patients with atrial fibrillation and multi-morbidity: Insights from the ARISTOTLE trial. Am. Heart J. (2019) doi: 10.1016/j.ahj.2018.09.017 . Al-Halawani, R., Qassem, M. & Kyriacou, P. A. Monte Carlo simulation of the effect of melanin concentration on light-tissue interactions in transmittance and reflectance finger photoplethysmography. Sci. Rep. 14, 1–17 (2024). Ajmal, Boonya-Ananta, T., Rodriguez, A. J., Du Le, V. N. & Ramella-Roman, J. C. Monte Carlo analysis of optical heart rate sensors in commercial wearables: the effect of skin tone and obesity on the photoplethysmography (PPG) signal. Biomed. Opt. Express 12, 7445 (2021). Rodriguez, A. J. et al. Skin optical properties in the obese and their relation to body mass index: a review. J. Biomed. Opt. 27, (2022). Boonya-ananta, T. et al. Synthetic photoplethysmography (PPG) of the radial artery through parallelized Monte Carlo and its correlation to body mass index (BMI). Sci. Rep. 11, 1–11 (2021). Szołtysek-Bołdys, I., Zielińska-Danch, W., Łoboda, D., Gołba, K. S. & Sarecka-Hujar, B. Do Photopletysmographic Parameters of Arterial Stiffness Differ Depending on the Presence of Arterial Hypertension and/or Atherosclerosis? Sensors 24, (2024). Insulander, P., Carnlöf, C., Schenck-Gustafsson, K. & Jensen-Urstad, M. Device profile of the Coala Heart Monitor for remote monitoring of the heart rhythm: overview of its efficacy. Expert Rev. Med. Devices (2020) doi: 10.1080/17434440.2020.1732814 . Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 20 Nov, 2025 Read the published version in npj Digital Medicine → Version 1 posted Editorial decision: Revision requested 29 Jun, 2025 Reviews received at journal 24 Jun, 2025 Reviews received at journal 23 Jun, 2025 Reviewers agreed at journal 17 Jun, 2025 Reviewers agreed at journal 16 Jun, 2025 Reviewers invited by journal 16 Jun, 2025 Editor assigned by journal 11 Jun, 2025 Submission checks completed at journal 11 Jun, 2025 First submitted to journal 08 Jun, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6849469","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":472048059,"identity":"e8680c50-efaf-469a-b51c-5d78ca8aee97","order_by":0,"name":"John Sollee","email":"","orcid":"","institution":"Northwestern University","correspondingAuthor":false,"prefix":"","firstName":"John","middleName":"","lastName":"Sollee","suffix":""},{"id":472048060,"identity":"9e828b18-bbd8-4766-98db-b0389934974d","order_by":1,"name":"Baljash Cheema","email":"","orcid":"","institution":"Northwestern University","correspondingAuthor":false,"prefix":"","firstName":"Baljash","middleName":"","lastName":"Cheema","suffix":""},{"id":472048061,"identity":"996deecb-4968-4083-a0fb-193058af9cc1","order_by":2,"name":"David Slotwiner","email":"","orcid":"","institution":"New York Presbyterian Queens, Weill Cornell Medical College","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"","lastName":"Slotwiner","suffix":""},{"id":472048062,"identity":"523b67ed-4107-487f-8034-b1568584ca2a","order_by":3,"name":"Alexander Volodarskiy","email":"","orcid":"","institution":"New York Presbyterian Queens, Weill Cornell Medical College","correspondingAuthor":false,"prefix":"","firstName":"Alexander","middleName":"","lastName":"Volodarskiy","suffix":""},{"id":472048063,"identity":"34eeea33-eb49-46cb-b04d-f5f1fa95e5a8","order_by":4,"name":"Lien Desteghe","email":"","orcid":"","institution":"University of Antwerp","correspondingAuthor":false,"prefix":"","firstName":"Lien","middleName":"","lastName":"Desteghe","suffix":""},{"id":472048064,"identity":"9158e02a-78b6-474c-a7d0-f89cffceabf7","order_by":5,"name":"Christophe Buyck","email":"","orcid":"","institution":"Antwerp University Hospital","correspondingAuthor":false,"prefix":"","firstName":"Christophe","middleName":"","lastName":"Buyck","suffix":""},{"id":472048065,"identity":"61a54892-e666-428a-8108-e5cc5ecebd97","order_by":6,"name":"Hein Heidbuchel","email":"","orcid":"","institution":"Hasselt University","correspondingAuthor":false,"prefix":"","firstName":"Hein","middleName":"","lastName":"Heidbuchel","suffix":""},{"id":472048066,"identity":"c5f221af-db29-435f-93b5-d921cf3aea64","order_by":7,"name":"Stavros Stavrakis","email":"","orcid":"","institution":"University of Oklahoma Health Sciences Center","correspondingAuthor":false,"prefix":"","firstName":"Stavros","middleName":"","lastName":"Stavrakis","suffix":""},{"id":472048067,"identity":"e27f5d98-83d8-4139-8119-7fbabfb8739a","order_by":8,"name":"Laurent Pison","email":"","orcid":"","institution":"Hospital Oost Limburg","correspondingAuthor":false,"prefix":"","firstName":"Laurent","middleName":"","lastName":"Pison","suffix":""},{"id":472048068,"identity":"674ea80e-2366-41fa-8a41-16bd887c5b9f","order_by":9,"name":"Dieter Nuyens","email":"","orcid":"","institution":"Hospital Oost Limburg","correspondingAuthor":false,"prefix":"","firstName":"Dieter","middleName":"","lastName":"Nuyens","suffix":""},{"id":472048069,"identity":"10defe44-1629-4aba-8f74-28effc152795","order_by":10,"name":"Maximo Rivero-Ayerza","email":"","orcid":"","institution":"Hospital Oost Limburg","correspondingAuthor":false,"prefix":"","firstName":"Maximo","middleName":"","lastName":"Rivero-Ayerza","suffix":""},{"id":472048070,"identity":"fa232f56-3656-4342-b33e-ecaafcde9a41","order_by":11,"name":"Hugo Herendael","email":"","orcid":"","institution":"Hospital Oost Limburg","correspondingAuthor":false,"prefix":"","firstName":"Hugo","middleName":"","lastName":"Herendael","suffix":""},{"id":472048077,"identity":"bac82f77-3bdc-42a5-9c60-c119654cfeb6","order_by":12,"name":"James D. Thomas","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA5ElEQVRIiWNgGAWjYBAC9gYw9T8BTH2oIEILI0SLBFgL44wzpGph5m0jRkv76cSHPxgk8vjFDj/+wDvvTjQDe+/jF3i19ORuNuZhkCiWnJ1mJiG57VluA89xMwv8DsvdJg10WOKG2wlmDIbbDuc2SKSxGeDV0v92+88fYC3pnz8kziFCi+CM3G0MPGAtOQYSBxvAWpgf4NMiLfF2szSPgUTizNk5ZZINxw7ntvEcY8Ong4GPP3fjxx8VEon90umbP/+pOZzbz97G/AGvHjBAdjvQCjYJwlrQADG2jIJRMApGwQgCAAzqS9d9Imr4AAAAAElFTkSuQmCC","orcid":"","institution":"Northwestern University","correspondingAuthor":true,"prefix":"","firstName":"James","middleName":"D.","lastName":"Thomas","suffix":""}],"badges":[],"createdAt":"2025-06-08 23:08:23","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6849469/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6849469/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41746-025-02059-2","type":"published","date":"2025-11-20T15:59:13+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":85361747,"identity":"c6b8655a-7c50-4f8a-af40-31173c2ad6b1","added_by":"auto","created_at":"2025-06-25 06:11:15","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":203666,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eData Acquisition\u003c/strong\u003e. Participants were instructed to sequentially place their fingers on the camera of each of the 10 devices to allow for 60 second PPG reading. A 12-lead electrocardiogram (ECG) was used as the reference standard. ECGs were performed for each participant twice during PPG recordings, once during the third smartphone PPG recording (iPhone 15) and once during the eighth smartphone PPG recording (Samsung Galaxy A53).\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6849469/v1/8f6dd66c746f1182b87470a3.jpeg"},{"id":96651041,"identity":"680cfdb5-4940-4a2b-8d68-9e0dc0dc8cd0","added_by":"auto","created_at":"2025-11-24 16:13:33","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1222930,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6849469/v1/f0980a57-6dc3-4f6c-b548-b0ad08c3923d.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"FibriCheck Detection Capabilities for Atrial Fibrillation (FDA – AF): A Multicenter Validation Study","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eAtrial fibrillation (AF) is the most common arrhythmia worldwide, which has significantly increased in both incidence and prevalence over the last 50 years, reaching the level of a cardiovascular disease (CVD) epidemic in the 21st century.\u003csup\u003e\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e Rising AF burden is driven by population aging\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e and increased rates of multimorbidity such as obesity, diabetes, hypertension (HTN), and chronic stress.\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e AF is causally associated with increased risk of myocardial infarction (MI), embolic ischemic stroke, heart failure (HF), and chronic kidney disease (CKD).\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e,\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e Development and progression of AF and its associated comorbidities are interdependent.\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e,\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e For instance, a sub-analysis of the Framingham Heart Study demonstrated that 37% of patients with new AF had HF, and conversely, 57% of patients with new HF had preexisting AF.\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e The prevalence of one condition was associated with a higher incidence of the other.\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e Patients with AF also have a 5 times increased risk of stroke, the leading cause of chronic severe disability in the US and the 5th leading cause of death.\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e Despite medical advances, AF often remains underdiagnosed, leading to preventable complications and mortality. Because AF is typically diagnosed by in-office 12-lead electrocardiogram (ECG), paroxysmal variants and asymptomatic cases are often missed.\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e Up to 20% of patients presenting with AF-related strokes are undiagnosed,\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e over 90% of whom meet criteria for chronic oral anticoagulation.\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e In short, there remains an ongoing need to develop clinically and economically feasible methods for early and accurate AF detection and monitoring to improve global public health.\u003c/p\u003e \u003cp\u003eWith recent advancements in microchips, sensor technologies, and cloud computing, researchers have developed a wide variety of tools for remote healthcare monitoring. Most recent innovations in the field of CVD and beyond have focused on consumer wearables or mobile smartphone applications, many of which incorporate artificial intelligence (AI).\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e Photoplethysmography (PPG) has emerged as the preferred signal modality for measuring heart rate and detecting arrhythmias, as smartphones and wearables already incorporate capable sensors.\u003csup\u003e\u003cspan additionalcitationids=\"CR15\" citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e In PPG, diodes emit light towards human tissue, and the reflected light is captured by photosensors. Because the intensity and pulsatility of the reflected light is a function of the propagation of arterial pressure pulses within the microvascular bed, PPG signals provide valuable real-time information about cardiovascular health, including oxygen saturation, heart rate, blood pressure, and cardiac output.\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e PPG-based AF detection algorithms typically work by extracting temporal, morphological, and/or spectral features from raw PPG signals, which are subsequently input into a classifier.\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e A recent review article identified 24 studies incorporating PPG for AF detection, half of which incorporated either machine or deep learning techniques.\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eFibriCheck \u0026copy; [Qompium NV, Hasselt, Belgium] is a CE Class IIa and FDA-cleared medical analysis platform that uses an end-to-end algorithm to classify heart rhythms based on PPG signals collected using their smartphone application. Prior studies, many from the multicenter European TeleCheck-AF project, demonstrated high usability, compliance, and patient satisfaction ratings,\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e,\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e including in the primary care setting.\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e Small, and/or single-center studies (N\u0026thinsp;\u0026le;\u0026thinsp;300) have demonstrated excellent performance in AF detection,\u003csup\u003e\u003cspan additionalcitationids=\"CR22 CR23\" citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e including in real-world conditions,\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e with the most recent validation study showing 100% sensitivity, 98.9% specificity, and 99.2% accuracy across 122 participants at a single European center.\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e Larger, multicenter validation studies in more diverse populations including non-European participants have not been performed. The FDA-AF study aimed to validate FibriCheck in a large, multi-center and multi-national cohort on the ten most popular smartphone devices.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy Design and Data Acquisition\u003c/h2\u003e \u003cp\u003eThe study was performed across five independent, large academic medical centers in the United States (US) and Europe: University Hospital Antwerp (UZA), Belgium; Hospital Oost-Limburg Genk (ZOL), Belgium; University of Oklahoma Health Sciences Center, Oklahoma, US; Northwestern Medicine, Chicago, US; New York Presbyterian Queens, New York, US. The institutional review board of each institution independently approved the study, and the study followed all principles of the Declaration of Helsinki (7th edition, October 2013), per the International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use \u0026ndash; Good Clinical Practice (ICH-GCP) guidelines. Written informed consent was obtained for all participants. Every attempt was made to protect patient confidentiality, and participants had the right to withdraw from the study at any time.\u003c/p\u003e \u003cp\u003eParticipants were eligible for inclusion if they met the following criteria: at least 22 years of age, capable of independently performing FibriCheck readings under observation from the study team and receiving active cardiology care either in the outpatient setting or hospitalized with or without AF. Enrollment was expected to last one month and attain 50% AF prevalence with 15\u0026ndash;20% class IV or higher on the Fitzpatrick scale to adequately sample patients with darker skin tone (a known population where PPG signal interpretation can be less accurate).\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e Participants were excluded if they had implantable pacemakers, cardioverter-defibrillators, or other electric devices, as such devices could interfere with natural heart rhythm; the model was not trained on such cases, and the FibriCheck platform is not intended for use in such patients. Participants were also excluded if they were unable to complete measurements independently due to physical or medical constraints, actively enrolled in other clinical trials, and/or pregnant or nursing women.\u003c/p\u003e \u003cp\u003eData acquisition for consented patients was performed at outpatient cardiology clinics or in hospital if patients were admitted. For each participant, the following demographic and clinical data were recorded: US or non-US citizen, BMI\u0026thinsp;\u0026ge;\u0026thinsp;30 or \u0026lt;\u0026thinsp;30, skin type (based on the Fitzpatrick scale), presence or absence of a history of AF, HF, vascular disease, hypertension (HTN), diabetes, and stroke.\u003c/p\u003e \u003cp\u003eThe FibriCheck application with access to the FibriCheck cloud and FibriCheck algorithm (v1.5.2) was pre-installed on 10 different iOS (n\u0026thinsp;=\u0026thinsp;8) and Android (n\u0026thinsp;=\u0026thinsp;2) devices. Participants were instructed to sequentially place their fingers on the camera of each of the 10 devices to allow for 60-second PPG reading as outlined in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The number of recording attempts per smartphone device was determined by the FibriCheck quality analysis, described below. If a PPG recording was deemed to have insufficient quality according to the FibriCheck algorithm, participants were instructed to repeat the recording until adequate quality was achieved, with a maximum of three attempts permitted per smartphone. At the end of each recording, the PPG data was automatically transmitted to the FibriCheck Cloud for processing and analysis by the FibriCheck algorithm.\u003c/p\u003e \u003cp\u003eA 12-lead electrocardiogram (ECG) was used as the reference standard. ECGs were performed for each participant twice during PPG recordings, once during the third smartphone PPG recording (Apple iPhone 15) and once during the eighth smartphone PPG recording (Samsung Galaxy A53) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Each ECG was evaluated by at least two board certified and independent cardiac electrophysiologists and labeled as regular, AF, atrial flutter, or unclassified (not one of the three other rhythms). If there were discrepancies in the findings of the two experts, then a third expert was consulted, and the majority decision decided the reference diagnosis. In the case of disagreement among all three experts or if the ECG was deemed unreadable (poor quality), the data were excluded from the analysis. If the two ECGs for a single patient were deemed to be different rhythms, then the data were also excluded.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eFibriCheck Platform Algorithm\u003c/h3\u003e\n\u003cp\u003eThe FibriCheck platform uses an end-to-end algorithm incorporating three convolutional neural networks (CNN): (1) \u003cem\u003equality detection\u003c/em\u003e, (2) \u003cem\u003eheartbeat detection\u003c/em\u003e, and (3) \u003cem\u003erhythm classification\u003c/em\u003e.\u003c/p\u003e \u003cp\u003eTo initiate data collection, users place their finger on the smartphone camera lens. Once the presence of a finger is confirmed using a dedicated detection algorithm, a video is recorded in YUV color format for one minute. After recording, the video is converted to the RGB color format, where the RGB components are treated as potential PPG signals. The RGB signals then undergo a series of preprocessing steps, which include noise filtering, derivative calculation, normalization, and signal truncation. These preprocessing steps are aimed at enhancing the quality of the extracted PPG signals and reducing the influence of noise or artifacts.\u003c/p\u003e \u003cp\u003eFollowing preprocessing, the (1) \u003cem\u003equality detection\u003c/em\u003e CNN indicates if specific segments within the PPG are too noisy for further analysis. If the model determines that more than 30 seconds of the PPG signal is too noisy or fails to meet quality standards, the measurement is flagged as \"insufficient quality,\" and no further clinical analysis is performed. Sufficient quality signals are passed to the (2) \u003cem\u003eheartbeat detection\u003c/em\u003e CNN, which indicates the location of heartbeats in the preprocessed PPG signal and constructs a PPG-based tachogram and average heart rate measurement over the one-minute measurement. Both the (1) \u003cem\u003equality detection\u003c/em\u003e and (2) \u003cem\u003eheartbeat detection\u003c/em\u003e models are trained on beat-to-beat annotated internal PPG datasets consisting of tens of thousands of synchronized PPG-ECG data samples.\u003c/p\u003e \u003cp\u003eWhen the signal quality meets the required criteria and the heart rate falls within the validated range, the platform proceeds to the (3) \u003cem\u003erhythm classification\u003c/em\u003e CNN. This model has been trained on a diverse dataset of over one million rhythm-annotated measurements, encompassing various heart rhythm disorders, including AF and atrial flutter amongst others. The algorithm then classifies the heart rhythm as regular, possible AF, or unclassified rhythm.\u003c/p\u003e\n\u003ch3\u003eTechnician Verification\u003c/h3\u003e\n\u003cp\u003eIn addition to fully automated classification by the \u003cem\u003erhythm classification\u003c/em\u003e CNN as above, each PPG recording of sufficient quality is independently reviewed by a blinded FibriCheck technician. Verification takes place within 48 hours and happens on a per-measurement basis: each measurement is first analyzed by the algorithm and then automatically queued for human verification. By visually reviewing the PPG recordings, the technician classifies the rhythm as \u0026ldquo;regular,\u0026rdquo; \u0026ldquo;possible AF,\u0026rdquo; or \u0026ldquo;unclassified rhythm\u0026rdquo; based on criteria extrapolated from the peer-reviewed practical guidance on signal interpretation and clinical scenarios from TeleCheck-AF.\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e Criteria for a regular rhythm include equal intervals between peaks in the raw PPG signal, sporadic irregularity, a single line or wave-like pattern in the tachogram, and a dense or ellipse-shaped cluster in the Poincar\u0026eacute; plot. Criteria for possible AF are irregularly varying intervals between the peaks in the PPG tracing, randomly distributed points on the tachogram, and the absence of a distinct cluster of points on the Poincar\u0026eacute; plot. If the recording does not meet criteria for either regular rhythm or possible AF, then it is labeled an unclassified rhythm. In the case of discrepancies between the automated and human classification, the automated classification is overruled, and the final classification is adjusted accordingly.\u003c/p\u003e\n\u003ch3\u003eStatistical Analysis and Outcome Measures\u003c/h3\u003e\n\u003cp\u003eSample size calculations were performed prior to subject recruitment. Based on European post-market surveillance data, an accuracy of 95.8% (95% CI: 93.76\u0026ndash;97.32%) can be expected in detecting AF. For the analysis, the reference value (p0) was set to the lower bound of the accuracy CI, which was 93.8%. The expected accuracy of the FibriCheck system was set at p\u0026thinsp;=\u0026thinsp;0.990 based on European data. Using the exact Clopper \u0026amp; Pearson method and aiming for a significance level (alpha) of 0.025 and a power (1-beta) of 0.8, the calculated sample size required for this comparison was determined to be 114 recordings of sufficient signal quality. To allow for potential sub analyses, the targeted number of participants was set at 250.\u003c/p\u003e \u003cp\u003eThe results obtained from the automated analysis by the FibriCheck system were compared with the 12-lead ECG reference diagnosis to evaluate the ability of the system to correctly differentiate the heart rhythm. Continuous variables were presented as mean with standard deviation (SD) or median with interquartile range (IQR). Categorical values were reported as counts and percentages. The performance of FibriCheck was evaluated by calculating the accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) along with corresponding 95% CI. Prevalence-adjusted PPV and NPV were also calculated assuming an AF prevalence of 6%. Data was analyzed using MedCalc Software Ltd.\u0026rsquo;s Diagnostic Test Evaluation Calculator (Version 22.016). The main outcome was overall performance across all devices and participants. Sub-group analyses were conducted to determine performance in individual devices, across different skin tones, in those with BMI over and under 30, and in those with and without a previous AF, HF, HTN, diabetes, vascular disease, or stroke diagnosis.\u003c/p\u003e\n\u003ch3\u003eComparative Analysis with FDA-Cleared Devices\u003c/h3\u003e\n\u003cp\u003eTo benchmark FibriCheck's performance, a comparative analysis was conducted using the FDA's 510(k) premarket notification database (Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). The analysis included 22 devices with similar indications for use, focusing on self-testing by patients diagnosed with or susceptible to AF. Among these, seven devices reported clinical testing results.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eStudy Population\u003c/h2\u003e \u003cp\u003eA total of 252 participants were enrolled. Of these, 16 (6.4%) were excluded from the study, including 4 who met exclusion criteria but were inappropriately enrolled, 5 dropouts due to withdrawn informed consent, 3 due to unavailability of a 12-lead ECG device during data acquisition, 2 due to interruption of the study by other medical examinations, 1 due to poor quality ECG, and 1 due to different rhythms being detected on the two ECG recordings for that single patient. Therefore, a total of 236 participants were included in the analysis. Demographics and clinical information are shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Among the 236 participants, 157 (66.5%) were identified as having regular rhythm, and 60 (25.4%) were identified as having AF based on 12-lead ECG (reference diagnosis). Atrial flutter was identified in 10 (4.2%) participants, and the remaining 9 (3.8%) participants presented with unclassified rhythms. For the primary analysis, atrial flutter recordings were excluded (99 recordings). Additionally, 48 recordings were flagged by the quality assessment algorithm and excluded prior to heartbeat detection and rhythm classification. Therefore, a total of 2,195 recordings were available for the primary analysis. Median time from PPG recording to capture of ECG reference diagnosis was 2 minutes (IQR 2,3).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eParticipant Demographics and Clinical Characteristics.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal population\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;236)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUS population\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;158)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNon-US population\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;78)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003eAge (in years)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian (Q1-Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e65 (54\u0026ndash;74)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e65 (50\u0026ndash;75)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e66 (58.8\u0026ndash;72)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSex, n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e143 (60.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e86 (54.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e57 (73.1%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e93 (39.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e72 (45.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e21 (26.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eBody Mass Index (kg/m\u003c/b\u003e\u003csup\u003e\u003cb\u003e2\u003c/b\u003e\u003c/sup\u003e\u003cb\u003e)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian (Q1-Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e28 (25\u0026ndash;32)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e29 (25\u0026ndash;33)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e27 (24.8\u0026ndash;30)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSkin tone, Fitzpatrick Scale\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eType I, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e47 (19.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e21 (13.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e26 (33.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eType II, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e113 (47.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e71 (44.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e42 (53.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eType III, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e28 (11.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e20 (12.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e8 (10.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eType IV, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e16 (6.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e15 (9.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1 (1.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eType V, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e15 (6.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e14 (8.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1 (1.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eType VI, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e17 (7.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e17 (10.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eMedical history\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAtrial fibrillation, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e148 (62.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e86 (54.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e62 (79.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePersistent atrial fibrillation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e68 (28.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e31 (19.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e25 (32.1%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eParoxysmal atrial fibrillation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e80 (33.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e55 (34.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e37 (47.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHeart failure, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e67 (28.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e59 (37.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e8 (10.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVascular disease, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e27 (11.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e22 (13.9%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5 (6.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHypertension, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e125 (53.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99 (62.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e26 (33.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiabetes, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e53 (22.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e42 (26.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e11 (14.1%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStroke, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e36 (15.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e27 (17.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9 (11.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCOPD diagnosis, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e15 (6.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10 (6.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5 (6.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003ePrimary Analysis: Overall Performance and Performance by Smartphone Device\u003c/h3\u003e\n\u003cp\u003eThe FibriCheck algorithm demonstrated high overall accuracy and reliability in differentiating AF from non-AF (Tables\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and \u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Without technician verification, accuracy was 98.5% (95% CI: 98.0%-99.0%), sensitivity was 96.3% (95% CI: 94.4%-97.7%), specificity was 99.3% (95% CI: 98.8%-99.7%), PPV was 98.0% (95% CI: 96.5%-98.9%), and NPV was 99.8% (95% CI: 99.6%-99.9%). Performance was not significantly changed with technician verification (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The FibriCheck algorithm demonstrated high performance in all 10 smartphone devices, with highest accuracy in the iPhone 13 Pro (100%) but no significant differences among all devices. Detailed performance by smartphone device is shown in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eFibriCheck Overall Performance Metrics.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePerformance Metric\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFibriCheck without Verification\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFibriCheck with Verification\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAccuracy (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.5% (98.0%-99.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e98.8% (98.2%-99.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSensitivity (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e96.3% (94.4%-97.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99.0% (97.7%-99.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpecificity (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e99.3% (98.8%-99.7%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e98.7% (98.0%-99.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDisease prevalence\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e26.1%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e26.2%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePPV (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.0% (96.5%-98.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e96.4% (94.7%-97.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrevalence-adjusted PPV, 6%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e90.1% (83.4%-94.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e82.9% (76.1%-88.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNPV (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.7% (98.1%-99.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99.6% (99.2%-99.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrevalence-adjusted NPV, 6%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e99.8% (99.6%-99.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99.9% (99.9%-100%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eFibriCheck Performance by Smartphone Device\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSmartphone Device\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAccuracy Possible AF vs non-possible AF\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eOverall performance (95% CI)\u003c/b\u003e\u003c/p\u003e \u003cp\u003eEligible for rhythm analysis*, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.5% (98.0%-99.0%)\u003c/p\u003e \u003cp\u003e2,195/2,243 (98.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eiPhone SE (3rd gen) (95% CI)\u003c/p\u003e \u003cp\u003eΔiPhone SE (3rd gen) - overall performance\u003c/p\u003e \u003cp\u003e Eligible for rhythm analysis*, n (%)\u003c/p\u003e \u003cp\u003eTime until reference diagnosis in minutes, median (Q1-Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e99.1% (96.8%-99.9%)\u003c/p\u003e \u003cp\u003e\u0026thinsp;+\u0026thinsp;0.6%\u003c/p\u003e \u003cp\u003e224/226 (99.1%)\u003c/p\u003e \u003cp\u003e6 (4,9)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eiPhone 15 Pro (95% CI)\u003c/p\u003e \u003cp\u003eΔiPhone 15 Pro - overall performance\u003c/p\u003e \u003cp\u003e Eligible for rhythm analysis*, n (%)\u003c/p\u003e \u003cp\u003eTime until reference diagnosis in minutes, median (Q1-Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.2% (95.4%-99.5%)\u003c/p\u003e \u003cp\u003e-0.3%\u003c/p\u003e \u003cp\u003e218/225 (96.9%)\u003c/p\u003e \u003cp\u003e5 (2,7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eiPhone 15 (95% CI)\u003c/p\u003e \u003cp\u003eΔiPhone 15 - overall performance\u003c/p\u003e \u003cp\u003e Eligible for rhythm analysis*, n (%)\u003c/p\u003e \u003cp\u003eTime until reference diagnosis in minutes, median (Q1-Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e96.8% (93.6%-98.7%)\u003c/p\u003e \u003cp\u003e-1.7%\u003c/p\u003e \u003cp\u003e221/226 (97.8%)\u003c/p\u003e \u003cp\u003eN/A, simultaneous\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eiPhone 11 (95% CI)\u003c/p\u003e \u003cp\u003eΔiPhone 11 - overall performance\u003c/p\u003e \u003cp\u003e Eligible for rhythm analysis*, n (%)\u003c/p\u003e \u003cp\u003eTime until reference diagnosis in minutes, median (Q1-Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.2% (95.5%-99.5%)\u003c/p\u003e \u003cp\u003e-0.3%\u003c/p\u003e \u003cp\u003e224/226 (99.1%)\u003c/p\u003e \u003cp\u003e3 (3,4)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eiPhone 14 (95% CI)\u003c/p\u003e \u003cp\u003eΔiPhone 14 - overall performance\u003c/p\u003e \u003cp\u003e Eligible for rhythm analysis*, n (%)\u003c/p\u003e \u003cp\u003eTime until reference diagnosis in minutes, median (Q1-Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e99.6% (97.5%-100%)\u003c/p\u003e \u003cp\u003e+\u0026thinsp;1.1%\u003c/p\u003e \u003cp\u003e221/225 (98.2%)\u003c/p\u003e \u003cp\u003e4 (3,5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eiPhone 12 (95% CI)\u003c/p\u003e \u003cp\u003eΔiPhone 12 - overall performance\u003c/p\u003e \u003cp\u003e Eligible for rhythm analysis*, n (%)\u003c/p\u003e \u003cp\u003eTime until reference diagnosis in minutes, median (Q1-Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.7% (96.1%-99.7%)\u003c/p\u003e \u003cp\u003e+\u0026thinsp;0.2%\u003c/p\u003e \u003cp\u003e223/225 (99.1%)\u003c/p\u003e \u003cp\u003e2 (1,3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSamsung Galaxy S23 (95% CI)\u003c/p\u003e \u003cp\u003eΔSamsung Galaxy S23 - overall performance\u003c/p\u003e \u003cp\u003e Eligible for rhythm analysis*, n (%)\u003c/p\u003e \u003cp\u003eTime until reference diagnosis in minutes, median (Q1-Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e97.2% (94.0%-99.0%)\u003c/p\u003e \u003cp\u003e-1.3%\u003c/p\u003e \u003cp\u003e213/221(96.4%)\u003c/p\u003e \u003cp\u003e2 (2,3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSamsung Galaxy A54 (95% CI)\u003c/p\u003e \u003cp\u003eΔSamsung Galaxy A54 - overall performance\u003c/p\u003e \u003cp\u003e Eligible for rhythm analysis*, n (%)\u003c/p\u003e \u003cp\u003eTime until reference diagnosis in minutes, median (Q1-Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.2% (95.4%-99.5%)\u003c/p\u003e \u003cp\u003e-0.3%\u003c/p\u003e \u003cp\u003e218/223 (97.8%)\u003c/p\u003e \u003cp\u003eN/A, simultaneous\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eiPhone 13 (95% CI)\u003c/p\u003e \u003cp\u003eΔiPhone 13 - overall performance\u003c/p\u003e \u003cp\u003e Eligible for rhythm analysis*, n (%)\u003c/p\u003e \u003cp\u003eTime until reference diagnosis in minutes, median (Q1-Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e99.5% (97.5%-100%)\u003c/p\u003e \u003cp\u003e+\u0026thinsp;1.0%\u003c/p\u003e \u003cp\u003e217/222 (97.7%)\u003c/p\u003e \u003cp\u003e2 (2,3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eiPhone 13 Pro (95% CI)\u003c/p\u003e \u003cp\u003eΔiPhone 13 Pro - overall performance\u003c/p\u003e \u003cp\u003e Eligible for rhythm analysis*, n (%)\u003c/p\u003e \u003cp\u003eTime until reference diagnosis in minutes, median (Q1-Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e100% (98.3%-100%)\u003c/p\u003e \u003cp\u003e+\u0026thinsp;1.5%\u003c/p\u003e \u003cp\u003e216/224 (96.4%)\u003c/p\u003e \u003cp\u003e3 (3,4)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eSubgroup Analyses\u003c/h2\u003e \u003cp\u003eSummary of performance in each subgroup is shown in Tables\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e and \u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e. The number of recordings that were excluded due to atrial flutter or poor quality is also shown in the table. The FibriCheck algorithm demonstrated high accuracy across all skin tones, although sensitivity was lower in participants with darker skin tones (Fitzpatrick type IV or higher). Verification by a FibriCheck technician successfully mitigated the risks related to the lower sensitivity of 79.6% (65.7% \u0026minus;\u0026thinsp;89.8%) in participants with a dark skin tone compared to participants with a pale or medium skin tone. The FibriCheck algorithm also showed high accuracy and reliability in participants with a previous AF diagnosis. The system performed consistently well in participants with and without HF, vascular disease, HTN, diabetes, and stroke.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSub-group Analysis Performance without Technician Verification.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"18\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c12\" colnum=\"12\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c13\" colnum=\"13\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c14\" colnum=\"14\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c15\" colnum=\"15\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c16\" colnum=\"16\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c17\" colnum=\"17\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c18\" colnum=\"18\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"17\" nameend=\"c18\" namest=\"c2\"\u003e \u003cp\u003ePossible AF vs non-possible AF\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSub-group\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBMI \u003c/p\u003e \u003cp\u003e\u0026ge;\u0026thinsp;30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eBMI \u003c/p\u003e \u003cp\u003e\u0026lt;\u0026thinsp;30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePale skin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMedium skin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eDark skin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eDiabetes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003eNo diabetes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003eAF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003cp\u003eAF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003eHF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003cp\u003eHF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003eStroke\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003cp\u003eStroke\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e \u003cp\u003eHTN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003eNo HTN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e \u003cp\u003eVascular disease\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e \u003cp\u003eNo vascular disease\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAccuracy \u003c/p\u003e \u003cp\u003e(95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e97.2% (95.9%-98.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99.5% (98.9%-99.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e98.7% (98.0%-99.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e99.5% (98.3%-99.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e96.3% (93.4%-98.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e97.8% (96.0%-98.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e98.8% (98.1%-99.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e97.9% (96.9%-98.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e\u003cb\u003e-\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e97.5% (95.9%-98.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e98.9% (98.3%-99.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e97.6% (95.4%-99.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e \u003cp\u003e98.7% (98.1%-99.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e \u003cp\u003e98.1% (97.2%-98.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e99.0% (98.2%-99.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e \u003cp\u003e100% (98.5%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e \u003cp\u003e98.4% (97.7%-98.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSensitivity (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e93.7% (89.9%-96.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e98.2% (96.1%-99.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e97.6% (95.7%-98.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e100% (94.7%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e79.6% (65.7%-89.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e93.6% (88.8%-96.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e97.5% (95.5%-98.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e96.3% (94.4%-97.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e96.0% (92.3%-98.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e96.5% (94.1%-98.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e89.9% (81.0%-95.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e \u003cp\u003e97.4% (95.5%-98.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e \u003cp\u003e96.1% (93.6%-97.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e96.8% (93.2%-98.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e \u003cp\u003e​​100% (95.1%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e \u003cp\u003e95.8% (93.6%-97.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003cp\u003e(95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.5% (97.2%-99.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99.90% (99.4%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e99.2% (98.5%-99.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e99.4% (98.0%-99.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e99.6% (97.8%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e100% (98.9%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e99.2% (98.5%-99.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e99.0% (98.0%-99.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e99.6% (98.7%-99.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e98.3% (96.5%-99.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e99.7% (99.2%-99.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e100% (98.6%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e \u003cp\u003e99.2% (98.6%-99.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e \u003cp\u003e99.1% (98.2%-99.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e99.5% (98.8%-99.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e \u003cp\u003e100% (97.8%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e \u003cp\u003e99.3% (98.7%-99.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExcluded (atrial flutter, poor quality)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e44 (20, 24)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e103 (79, 24)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e106 (69, 37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e106 (69, 37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e106 (69, 37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e33 (20, 13)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e114 (79, 35)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e117 (79, 38)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e30 (20, 10)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e51 (40, 11)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e96 (59, 37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e132 (89, 43)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e \u003cp\u003e132 (89, 43)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e \u003cp\u003e67 (40, 27)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e80 (59, 21)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e \u003cp\u003e24 (20, 4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e \u003cp\u003e123 (79, 44)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSub-group Analysis Performance with Technician Verification.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"18\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c12\" colnum=\"12\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c13\" colnum=\"13\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c14\" colnum=\"14\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c15\" colnum=\"15\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c16\" colnum=\"16\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c17\" colnum=\"17\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c18\" colnum=\"18\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"17\" nameend=\"c18\" namest=\"c2\"\u003e \u003cp\u003ePossible AF vs non-possible AF\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSub-group\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBMI \u003c/p\u003e \u003cp\u003e\u0026ge;\u0026thinsp;30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eBMI \u003c/p\u003e \u003cp\u003e\u0026lt;\u0026thinsp;30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePale skin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMedium skin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eDark skin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eDiabetes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003eNo diabetes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003eAF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003cp\u003eAF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003eHF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003cp\u003eHF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003eStroke\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003cp\u003eStroke\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e \u003cp\u003eHTN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003eNo HTN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e \u003cp\u003eVascular disease\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e \u003cp\u003eNo vascular disease\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAccuracy \u003c/p\u003e \u003cp\u003e(95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.2% (97.1%-99.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99.2% (98.5%-99.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e98.7% (98.0%-99.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e99.0% (97.6%-99.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e98.6% (96.6%-99.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e99.0% (97.6%-99.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e98.7% (98.1%-99.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e98.5% (97.7%-99.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e\u003cb\u003e-\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e97.9% (96.4%-98.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e99.1% (98.5%-99.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e99.4% (97.9%-99.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e \u003cp\u003e98.7% (98.0%-99.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e \u003cp\u003e99.0% (98.2%-99.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e98.5% (97.6%-99.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e \u003cp\u003e99.2% (97.0%-99.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e \u003cp\u003e98.7% (98.1%-99.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003cp\u003e(95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.8% (96.4%-99.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99.1% (97.4%-99.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e99.3% (98.1%-99.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e100% (94.7%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e93.8% (82.8%-98.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e98.8% (95.8%-99.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e99.0% (97.5%-99.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e99.0% (97.7%-99.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e\u003cb\u003e-\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e98.0% (95.0%-99.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e99.5% (98.1%-99.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e97.4% (91.0%-99.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e \u003cp\u003e99.2% (98.0%-99.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e \u003cp\u003e99.2% (97.7%-99.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e98.4% (95.5%-99.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e \u003cp\u003e100% (95.1%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e \u003cp\u003e98.8% (97.4%-99.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003cp\u003e(95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.0% (96.6%-98.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99.2% (98.4%-99.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e98.4% (97.5%-99.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e98.9% (97.1%-99.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e99.6% (97.8%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e99.1% (97.3%-99.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e98.6% (97.8%-99.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e98.2% (97.0%-99.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e99.2% (98.3%-99.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e97.8% (95.8%-99.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e99.0%\u003c/p\u003e \u003cp\u003e(98.3%- 99.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e100% (98.6%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e \u003cp\u003e98.5% (97.6%-99.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e \u003cp\u003e98.9% (97.8%-99.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e98.6% (97.5%-99.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e \u003cp\u003e98.8% (95.7%-99.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e \u003cp\u003e98.7% (98.0%-99.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExcluded (atrial flutter, poor quality)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e44 (20, 24)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e103 (79, 24)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e106 (69, 37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e106 (69, 37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e106 (69, 37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e33 (20, 13)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e114 (79, 35)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e117 (79, 38)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e30 (20, 10)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e51 (40, 11)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e96 (59, 37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e132 (89, 43)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e \u003cp\u003e132 (89, 43)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e \u003cp\u003e67 (40, 27)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e80 (59, 21)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e \u003cp\u003e24 (20, 4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e \u003cp\u003e123 (79, 44)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eComparative Analysis with FDA-Cleared Devices\u003c/h2\u003e \u003cp\u003eCompared to the 22 devices with similar indications for use, FibriCheck demonstrated superior or equivocal sensitivity and specificity. Full resulted are shown in Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eOverview of reported performance metrics of previously cleared devices with a similar indication for use and reported clinical performance based on the 510(k) premarket notification database.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDevice name\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e510(k) number\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003cp\u003e(95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSpecificity \u003c/p\u003e \u003cp\u003e(95% CI)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eFibriCheck\u003c/span\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eK232804\u003c/span\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e96.3% \u003c/span\u003e\u003c/p\u003e \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e(94.4%-97.7%)\u003c/span\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e99.3% \u003c/span\u003e\u003c/p\u003e \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e(98.8%-99.7%)\u003c/span\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFibriCheck\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eK173872\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95.60%\u003c/p\u003e \u003cp\u003e(no 95% CI reported)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e96.6%\u003c/p\u003e \u003cp\u003e(no 95% CI reported)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCoala Heart Monitor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eK182040\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e97.2%\u003c/p\u003e \u003cp\u003e(no 95% CI reported)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e94.6%\u003c/p\u003e \u003cp\u003e(no 95% CI reported)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStudy Watch with Irregular Pulse Monitor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eK192415\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e85%\u003c/p\u003e \u003cp\u003e(79%-90%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e96%\u003c/p\u003e \u003cp\u003e(93%-99%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHalo AF Detection System\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eK201208\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e93.3%\u003c/p\u003e \u003cp\u003e(no 95% CI reported)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e99.1%\u003c/p\u003e \u003cp\u003e(no 95% CI reported)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eScan Monitor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eK201456\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e96.3%\u003c/p\u003e \u003cp\u003eLower bound 95% CI: 89.4%, No upper bound reported\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e100%\u003c/p\u003e \u003cp\u003eLower bound 95% CI: 96.7% No upper bound reported\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStudy Watch with Irregular Pulse Monitor (Home) Study Watch with Irregular Pulse Monitor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eK213357\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e96.1%\u003c/p\u003e \u003cp\u003e(92.7%-98.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e98.1%\u003c/p\u003e \u003cp\u003e(97.2%-99.1%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWithings Scan Monitor 2.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eK230812\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e99%\u003c/p\u003e \u003cp\u003e(93%-100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e99%\u003c/p\u003e \u003cp\u003e(97%-100%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eThe FDA-AF study validates the FibriCheck platform as a highly accurate and reliable tool for detecting AF in a diverse patient population. By demonstrating consistent performance across 10 of the most common smartphone devices, the study also underscores the platform\u0026rsquo;s ease of implementation and potential as a resource-efficient method for AF detection and monitoring outside of the clinical setting.\u003c/p\u003e \u003cp\u003eThe FibriCheck platform offers several clinical advantages over existing methods for AF detection and monitoring. Unlike traditional 12-lead ECGs, FibriCheck measurements can be performed at any time, within 60 seconds, and utilizing a device already owned by most patients. Patients without a formal diagnosis of AF but exhibiting symptoms may be instructed by a clinician to initiate FibriCheck readings when feeling symptomatic. Similarly, those with paroxysmal AF may be instructed to take periodic readings to assess AF burden. For select paroxysmal patients who are managed with a \u0026ldquo;pill in the pocket\u0026rdquo; approach, FibriCheck could guide self-administration of single dose antiarrhythmics (e.g., flecainide or propafenone) to promptly terminate the arrythmia.\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e FibriCheck can also be used to monitor for recurrence following electrical cardioversion or ablation procedures.\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e,\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eLike other patient-activated wearables including smartwatches and handheld ECG devices, FibriCheck may miss transient or asymptomatic arrythmias.\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e Still, unlike continuous monitoring devices such as the ZioPatch,\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e Holter monitor, or loop recorder, FibriCheck is entirely non-invasive, does not require external battery packs or chest leads, and can record and transmit unlimited readings without the need for repeat office visits or hardware exchanges.\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e Wrist-worn devices for continuous AF monitoring are being developed, such as the recently FDA-cleared Verily Study Watch;\u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e however, these require the purchase of additional hardware rather than operating through a basic smartphone. In contrast, FibriCheck operates on devices already widely available, making it particularly suitable for resource-limited settings\u003c/p\u003e \u003cp\u003eThe FibriCheck algorithm performed well across a diverse patient population, reinforcing clinical utility, particularly given high rates of multimorbidity in AF.\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e,\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e Several studies have demonstrated that certain comorbidities and patient characteristics, most notably obesity and skin tone, can significantly affect PPG signal quality and lead to inaccurate biophysical measurements. Skin tone is often described using the Fitzpatrick scale, which classifies skin types from I, the lightest, to VI, the darkest based on response to ultraviolet light.\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e Monte Carlo simulations have shown that the AC/DC ratio of PPG signals, a measure of blood volume pulsatility detection, is compromised in darker skin (higher Fitzpatrick scale) due to increased light absorption by melanin.\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e,\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e This effect has been shown to result in signal loss in existing commercial wearables, including the Apple Watch series 5 and Fitbit Versa 2.\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e Obesity also affects PPG signal quality due to the effects of adipose and dermal tissue on penetration and scattering of light,\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e with effects on AC/DC signal degradation up to 40%.\u003csup\u003e39\u003c/sup\u003e Vascular disease and HTN have also been shown to affect PPG signals, however likely to a lesser extent.\u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eThe subgroup analysis demonstrated that FibriCheck maintains high accuracy, sensitivity, and specificity in patients with diabetes and prior stroke as well as pre-existing diagnoses of HF, HTN, and vascular disease. Sensitivity was reduced in those with darker skin tone, but this was mitigated by FibriCheck technician verification; with verification, sensitivity improved from 79.6\u0026ndash;93.8%. Likewise, sensitivity was slightly reduced in those with BMI\u0026thinsp;\u0026ge;\u0026thinsp;30 kg/m\u003csup\u003e2\u003c/sup\u003e. This was also mitigated with technician verification, improving sensitivity from 93.7\u0026ndash;98.8%. By offering technician verification, the FibriCheck platform can successfully mitigate the known effects of skin type and obesity on classification performance. This feature affords FibriCheck an advantage in comparison to other consumer platforms for mobile AF detection that do not offer human verification.\u003c/p\u003e \u003cp\u003eTo benchmark FibriCheck to the state-of-the-art, we performed a comparative analysis based on performance metrics of previously cleared devices with a similar indication for use and reported clinical performance based on the 510(k) premarket notification database. FibriCheck demonstrated comparable or superior performance to all identified devices reporting performance metrics.\u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e,\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eIn conclusion, the FDA-AF study confirms the high accuracy, sensitivity, and specificity of the FibriCheck algorithm in detecting AF across various smartphone platforms and clinical subgroups. These findings support the use of FibriCheck as a reliable, low-cost, and easily accessible tool for AF detection in a diverse patient population.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eAF: atrial fibrillation; PPG: photoplethysmography; CI: confidence interval; HTN: hypertension; HF: heart failure; CKD: chronic kidney disease; AI: artificial intelligence; ECG: electrocardiogram; CNN: convolutional neural network.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003eConflict of Interest Statement:\u003c/h2\u003e \u003cp\u003eJT: Consulting with GE Healthcare, Caption Health, Abbott, Eko Health. BJ: Consulting fees from Caption Health, Inc. and Viz.ai; served on an advisory board for Novo Nordisk, is an advisor with equity in Healthspan, Inc. and Zoe Biosciences; received speaking fees and honoraria from Bristol Meyers Squibb. All others report no relevant conflicts of interest.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eWriting - Original Draft and Visualization: J.S. Writing - Review \u0026amp; Editing: B.C., J.T. Investigation, Data Curation, and Project Administration: B.C., J.T., D. S., A.V., L.D., C.B., H.H., S.S., L.P., D.N., M.R-A., H.V.H. Supervision: J.T.\u003c/p\u003e\u003ch2\u003eAcknowledgments:\u003c/h2\u003e \u003cp\u003eThe authors would like to thank the FibriCheck team at Qompium NV, Hasselt, Belgium.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eSchnabel, R. B. \u003cem\u003eet al.\u003c/em\u003e Fifty-Year Trends in Atrial Fibrillation Prevalence, Incidence, Risk Factors, and Mortality in the Community Renate. \u003cem\u003eLancet\u003c/em\u003e 386, 154\u0026ndash;162 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChugh, S. S. \u003cem\u003eet al.\u003c/em\u003e Worldwide epidemiology of atrial fibrillation: A global burden of disease 2010 study. \u003cem\u003eCirculation\u003c/em\u003e 129, 837\u0026ndash;847 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKornej, J., B\u0026ouml;rschel, C. S., Benjamin, E. J. \u0026amp; Schnabel, R. B. Epidemiology of Atrial Fibrillation in the 21st Century. \u003cem\u003eCirc. Res.\u003c/em\u003e 127, 4\u0026ndash;20 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHeidenreich, P. A. \u003cem\u003eet al.\u003c/em\u003e Forecasting the future of cardiovascular disease in the United States: A policy statement from the American Heart Association. \u003cem\u003eCirculation\u003c/em\u003e 123, 933\u0026ndash;944 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBisbal, F., Baranchuk, A., Braunwald, E., Bay\u0026eacute;s de Luna, A. \u0026amp; Bay\u0026eacute;s-Gen\u0026iacute;s, A. Atrial Failure as a Clinical Entity: JACC Review Topic of the Week. \u003cem\u003eJ. Am. Coll. Cardiol.\u003c/em\u003e 75, 222\u0026ndash;232 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBenjamin, E. J. \u003cem\u003eet al. Heart Disease and Stroke Statistics-2019 Update: A Report From the American Heart Association\u003c/em\u003e. \u003cem\u003eCirculation\u003c/em\u003e vol. 139 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBorowsky, L. H. \u003cem\u003eet al.\u003c/em\u003e First Diagnosis of Atrial Fibrillation at the Time of Stroke. \u003cem\u003eCerebrovasc. Dis.\u003c/em\u003e 43, 192\u0026ndash;199 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSanthanakrishnan, R. \u003cem\u003eet al.\u003c/em\u003e Atrial fibrillation begets heart failure and vice versa: Temporal associations and differences in preserved versus reduced ejection fraction. \u003cem\u003eCirculation\u003c/em\u003e (2016) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1161/CIRCULATIONAHA.115.018614\u003c/span\u003e\u003cspan address=\"10.1161/CIRCULATIONAHA.115.018614\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJanuary, C. T. \u003cem\u003eet al.\u003c/em\u003e 2014 \u003cem\u003eAHA/ACC/HRS guideline for the management of patients with atrial fibrillation: A report of the American College of cardiology/American heart association task force on practice guidelines and the heart rhythm society\u003c/em\u003e. \u003cem\u003eCirculation\u003c/em\u003e vol. 130 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSavelieva, I. \u0026amp; Camm, A. J. Clinical relevance of silent atrial fibrillation: Prevalence, prognosis, quality of life, and management. \u003cem\u003eJournal of Interventional Cardiac Electrophysiology\u003c/em\u003e (2000) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1023/A:1009823001707\u003c/span\u003e\u003cspan address=\"10.1023/A:1009823001707\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTurakhia, M. P. \u003cem\u003eet al.\u003c/em\u003e Contemporary prevalence estimates of undiagnosed and diagnosed atrial fibrillation in the United States. \u003cem\u003eClin. Cardiol.\u003c/em\u003e 46, 484\u0026ndash;493 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePetek, B. J. \u003cem\u003eet al.\u003c/em\u003e Consumer Wearable Health and Fitness Technology in Cardiovascular Medicine: JACC State-of-the-Art Review. \u003cem\u003eJ. Am. Coll. Cardiol.\u003c/em\u003e 82, 245\u0026ndash;264 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSpatz, E. S., Ginsburg, G. S., Rumsfeld, J. S. \u0026amp; Turakhia, M. P. Wearable Digital Health Technologies for Monitoring in Cardiovascular Medicine. \u003cem\u003eN. Engl. J. Med.\u003c/em\u003e (2024) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1056/nejmra2301903\u003c/span\u003e\u003cspan address=\"10.1056/nejmra2301903\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTang, S. C. \u003cem\u003eet al.\u003c/em\u003e Identification of Atrial Fibrillation by Quantitative Analyses of Fingertip Photoplethysmogram. \u003cem\u003eSci. Rep.\u003c/em\u003e 7, 1\u0026ndash;7 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePereira, T. \u003cem\u003eet al.\u003c/em\u003e Photoplethysmography based atrial fibrillation detection: a review. \u003cem\u003enpj Digit. Med.\u003c/em\u003e 3, (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eO\u0026rsquo;Sullivan, J. W. \u003cem\u003eet al.\u003c/em\u003e Accuracy of Smartphone Camera Applications for Detecting Atrial Fibrillation A Systematic Review and Meta-analysis. \u003cem\u003eJAMA Netw. Open\u003c/em\u003e 3, E202064 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlmarshad, M. A., Islam, M. S., Al-Ahmadi, S. \u0026amp; Bahammam, A. S. Diagnostic Features and Potential Applications of PPG Signal in Healthcare: A Systematic Review. \u003cem\u003eHealthc.\u003c/em\u003e 10, 1\u0026ndash;28 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGawałko, M. \u003cem\u003eet al.\u003c/em\u003e The European TeleCheck-AF project on remote app-based management of atrial fibrillation during the COVID-19 pandemic: Centre and patient experiences. \u003cem\u003eEuropace\u003c/em\u003e 23, 1003\u0026ndash;1015 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGawałko, M. \u003cem\u003eet al.\u003c/em\u003e Patient motivation and adherence to an on-demand app-based heart rate and rhythm monitoring for atrial fibrillation management: data from the TeleCheck-AF project. \u003cem\u003eEur. J. Cardiovasc. Nurs.\u003c/em\u003e 22, 412\u0026ndash;424 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBeerten, S. G., Proesmans, T. \u0026amp; Vaes, B. A heart rate monitoring app (fibricheck) for atrial fibrillation in general practice: Pilot usability study. \u003cem\u003eJMIR Form. Res.\u003c/em\u003e 5, 1\u0026ndash;9 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGruwez, H. \u003cem\u003eet al.\u003c/em\u003e Head-to-head comparison of proprietary PPG and single-lead ECG algorithms for atrial fibrillation detection. \u003cem\u003eEP Eur.\u003c/em\u003e 23, 2021 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSelder, J. L. \u003cem\u003eet al.\u003c/em\u003e Assessment of a standalone photoplethysmography (PPG) algorithm for detection of atrial fibrillation on wristband-derived data. \u003cem\u003eComput. Methods Programs Biomed\u003c/em\u003e. 197, (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eProesmans, T. \u003cem\u003eet al.\u003c/em\u003e Mobile phone\u0026ndash;based use of the photoplethysmography technique to detect atrial fibrillation in primary care: Diagnostic accuracy study of the fibricheck app. \u003cem\u003eJMIR mHealth uHealth\u003c/em\u003e (2019) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2196/12284\u003c/span\u003e\u003cspan address=\"10.2196/12284\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGruwez, H. \u003cem\u003eet al.\u003c/em\u003e Evaluation of the device independent nature of a photoplethysmography-deriving smartphone app. \u003cem\u003eEP Eur.\u003c/em\u003e 23, 2021 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGruwez, H. \u003cem\u003eet al.\u003c/em\u003e Real-world validation of smartphone-based photoplethysmography for rate and rhythm monitoring in atrial fibrillation. \u003cem\u003eEuropace\u003c/em\u003e 26, 1\u0026ndash;9 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWouters, F. \u003cem\u003eet al.\u003c/em\u003e Comparative Evaluation of Consumer Wearable Devices for Atrial Fibrillation Detection: Validation Study. \u003cem\u003eJMIR Form. Res.\u003c/em\u003e 9, 1\u0026ndash;11 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGupta, V. \u0026amp; Sharma, V. K. Skin typing: Fitzpatrick grading and others. \u003cem\u003eClin. Dermatol.\u003c/em\u003e (2019) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.clindermatol.2019.07.010\u003c/span\u003e\u003cspan address=\"10.1016/j.clindermatol.2019.07.010\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Der Velden, R. M. J. \u003cem\u003eet al.\u003c/em\u003e The photoplethysmography dictionary: Practical guidance on signal interpretation and clinical scenarios from TeleCheck-AF. \u003cem\u003eEur. Hear. J. - Digit. Heal.\u003c/em\u003e 2, 363\u0026ndash;373 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJoglar, J. A. \u003cem\u003eet al.\u003c/em\u003e 2023 \u003cem\u003eACC/AHA/ACCP/HRS Guideline for the Diagnosis and Management of Atrial Fibrillation: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines\u003c/em\u003e. \u003cem\u003eCirculation\u003c/em\u003e vol. 149 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCalvert, P. \u003cem\u003eet al.\u003c/em\u003e Remote rhythm monitoring using a photoplethysmography smartphone application after cardioversion for atrial fibrillation. \u003cem\u003eEur. Hear. J. - Digit. Heal.\u003c/em\u003e 5, 461\u0026ndash;468 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGruwez, H. \u003cem\u003eet al.\u003c/em\u003e Effectiveness of photoplethysmography-based rhythm monitoring after atrial fibrillation ablation using a smartphone application: DIGITOTAL study. \u003cem\u003eHear. Rhythm\u003c/em\u003e (2024) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.hrthm.2024.11.026\u003c/span\u003e\u003cspan address=\"10.1016/j.hrthm.2024.11.026\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTurakhia, M. P. \u003cem\u003eet al.\u003c/em\u003e Diagnostic utility of a novel leadless arrhythmia monitoring device. \u003cem\u003eAm. J. Cardiol.\u003c/em\u003e (2013) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.amjcard.2013.04.017\u003c/span\u003e\u003cspan address=\"10.1016/j.amjcard.2013.04.017\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePoh, M. Z. \u003cem\u003eet al.\u003c/em\u003e Validation of a Deep Learning Algorithm for Continuous, Real-Time Detection of Atrial Fibrillation Using a Wrist-Worn Device in an Ambulatory Environment. \u003cem\u003eJ. Am. Heart Assoc.\u003c/em\u003e 12, 1\u0026ndash;13 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArnett, D. K. \u003cem\u003eet al.\u003c/em\u003e AHA/ACC/HHS strategies to enhance application of clinical practice guidelines in patients with cardiovascular disease and comorbid conditions from the American heart association, American college of cardiology, and US department of health and human servic. \u003cem\u003eCirculation\u003c/em\u003e 130, 1662\u0026ndash;1667 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlexander, K. P. \u003cem\u003eet al.\u003c/em\u003e Outcomes of apixaban versus warfarin in patients with atrial fibrillation and multi-morbidity: Insights from the ARISTOTLE trial. \u003cem\u003eAm. Heart J.\u003c/em\u003e (2019) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.ahj.2018.09.017\u003c/span\u003e\u003cspan address=\"10.1016/j.ahj.2018.09.017\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAl-Halawani, R., Qassem, M. \u0026amp; Kyriacou, P. A. Monte Carlo simulation of the effect of melanin concentration on light-tissue interactions in transmittance and reflectance finger photoplethysmography. \u003cem\u003eSci. Rep.\u003c/em\u003e 14, 1\u0026ndash;17 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAjmal, Boonya-Ananta, T., Rodriguez, A. J., Du Le, V. N. \u0026amp; Ramella-Roman, J. C. Monte Carlo analysis of optical heart rate sensors in commercial wearables: the effect of skin tone and obesity on the photoplethysmography (PPG) signal. \u003cem\u003eBiomed. Opt. Express\u003c/em\u003e 12, 7445 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRodriguez, A. J. \u003cem\u003eet al.\u003c/em\u003e Skin optical properties in the obese and their relation to body mass index: a review. \u003cem\u003eJ. Biomed. Opt.\u003c/em\u003e 27, (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoonya-ananta, T. \u003cem\u003eet al.\u003c/em\u003e Synthetic photoplethysmography (PPG) of the radial artery through parallelized Monte Carlo and its correlation to body mass index (BMI). \u003cem\u003eSci. Rep.\u003c/em\u003e 11, 1\u0026ndash;11 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSzołtysek-Bołdys, I., Zielińska-Danch, W., Łoboda, D., Gołba, K. S. \u0026amp; Sarecka-Hujar, B. Do Photopletysmographic Parameters of Arterial Stiffness Differ Depending on the Presence of Arterial Hypertension and/or Atherosclerosis? \u003cem\u003eSensors\u003c/em\u003e 24, (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eInsulander, P., Carnl\u0026ouml;f, C., Schenck-Gustafsson, K. \u0026amp; Jensen-Urstad, M. Device profile of the Coala Heart Monitor for remote monitoring of the heart rhythm: overview of its efficacy. \u003cem\u003eExpert Rev. Med. Devices\u003c/em\u003e (2020) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/17434440.2020.1732814\u003c/span\u003e\u003cspan address=\"10.1080/17434440.2020.1732814\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"npj-digital-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjdigitalmed","sideBox":"Learn more about [npj Digital Medicine](http://www.nature.com/npjdigitalmed/)","snPcode":"41746","submissionUrl":"https://submission.springernature.com/new-submission/41746/3","title":"npj Digital Medicine","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"atrial fibrillation, artificial intelligence, consumer devices","lastPublishedDoi":"10.21203/rs.3.rs-6849469/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6849469/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eAtrial fibrillation (AF) is the most common arrhythmia worldwide and is associated with significant morbidity, mortality, and healthcare spending. Despite medical advances, AF remains underdiagnosed and undertreated, leading to preventable complications. FibriCheck \u0026copy; [Qompium NV, Hasselt, Belgium] is a medical analysis platform that uses an end-to-end algorithm to detect AF based on photoplethysmography (PPG) signals recorded on consumer smartphones.\u003c/p\u003e\u003ch2\u003ePurpose\u003c/h2\u003e \u003cp\u003eThe study aimed to validate FibriCheck in a large, multi-center and multi-national cohort on ten popular smartphone devices.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eA total of 236 patients were recruited from five independent, large academic centers in the United States and Europe. The FibriCheck system incorporates several convolutional neural networks to detect individual heartbeats, estimate average heart rate, and classify the rhythm based on PPG signals. Classification is verified by a FibriCheck technician. Classification performance was compared to the standard 12-lead electrocardiogram in the study population. Performance was assessed across clinical subgroups and smartphone devices.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eFibriCheck demonstrated high overall accuracy and reliability in detecting AF without technician verification: accuracy 98.5% (95% CI: 98.0%-99.0%); sensitivity 96.3% (95% CI: 94.4%-97.7%); specificity 99.3% (95% CI: 98.8%-99.7%); positive predictive value 98.0% (95% CI: 96.5%-98.9%); negative predictive value 99.8% (95% CI: 99.6%-99.9%). Performance was not affected by smartphone device or the presence or absence of comorbid heart failure, vascular disease, hypertension, diabetes, or stroke. Sensitivity was reduced in those with darker skin tone and higher BMI, but this was mitigated by technician verification.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eThe study confirms the high accuracy, sensitivity, and specificity of the FibriCheck algorithm in detecting AF across various smartphone models and clinical subgroups. These findings support the use of FibriCheck as a reliable, low-cost, and easily accessible tool for AF detection in a diverse patient population.\u003c/p\u003e","manuscriptTitle":"FibriCheck Detection Capabilities for Atrial Fibrillation (FDA – AF): A Multicenter Validation Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-25 06:11:09","doi":"10.21203/rs.3.rs-6849469/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-06-30T03:59:25+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-06-24T15:53:23+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-06-23T04:54:03+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"146782928935376819120355455188046386809","date":"2025-06-18T03:39:35+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"104153204192409165473993961891651530591","date":"2025-06-16T05:58:39+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-06-16T05:53:03+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-06-12T00:16:27+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-06-11T15:25:08+00:00","index":"","fulltext":""},{"type":"submitted","content":"npj Digital Medicine","date":"2025-06-08T23:03:13+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"npj-digital-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjdigitalmed","sideBox":"Learn more about [npj Digital Medicine](http://www.nature.com/npjdigitalmed/)","snPcode":"41746","submissionUrl":"https://submission.springernature.com/new-submission/41746/3","title":"npj Digital Medicine","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"fa306041-39b0-4994-af77-9f84672bdb64","owner":[],"postedDate":"June 25th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":50128884,"name":"Health sciences/Diseases/Cardiovascular diseases/Arrhythmias/Atrial fibrillation"},{"id":50128885,"name":"Biological sciences/Computational biology and bioinformatics/Machine learning"}],"tags":[],"updatedAt":"2025-11-24T16:10:51+00:00","versionOfRecord":{"articleIdentity":"rs-6849469","link":"https://doi.org/10.1038/s41746-025-02059-2","journal":{"identity":"npj-digital-medicine","isVorOnly":false,"title":"npj Digital Medicine"},"publishedOn":"2025-11-20 15:59:13","publishedOnDateReadable":"November 20th, 2025"},"versionCreatedAt":"2025-06-25 06:11:09","video":"","vorDoi":"10.1038/s41746-025-02059-2","vorDoiUrl":"https://doi.org/10.1038/s41746-025-02059-2","workflowStages":[]},"version":"v1","identity":"rs-6849469","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6849469","identity":"rs-6849469","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}