Heart_Rhythm is a free app that allows the user to record PPG and then compare the PPG with a model sinus rhythm or atrial fibrillation wave. The efficacy of such subjective self-assessments of rhythm patterns has not been validated in any clinical research [26].

Photo Afib Detector is a free app, which automatically detects an abnormality in the pattern of live-recorded PPG signal by estimating 2 statistical parameters: root mean square of successive difference and Shannon entropy [27]. An algorithm combining root mean square of successive difference and Shannon entropy in an iPhone 4S showed 96.2% sensitivity and 97.5% specificity for beat-to-beat discrimination of AF from sinus rhythm when compared with the 12-lead ECG [28]. However, Photo AFib Detector has not been directly validated in any clinical research.

Cardiio: Pulsometer is a free app, while Cardiio Rhythm is a beta version currently used only for scientific purposes. 

Cardiio: Pulsometer records high-quality PPG that can be evaluated by an expert and classified as sinus rhythm or rhythm other than sinus; unfortunately, there is no automatic algorithm for arrhythmia detection. Interestingly, the previous version of Cardiio: Pulsometer, called Cardiio – Heart Rate Monitor, was equipped with a face mode that enabled a contactless measurement of the heart rate based on the face of the user. Although Yan et al [29] showed that both finger and face PPGs demonstrate high accuracy in measuring resting heart rates, the app currently only uses the finger mode due to legal reasons [30].

Although Cardiio Rhythm is not currently commercially available, recent clinical findings regarding the app are promising. The sensitivity of the Cardiio Rhythm finger mode (92.9%, 95% CI 77-99) was found to be higher than the internet-enabled mobile ECG distributed by AliveCor (iECG; 71.4, 95% CI 51-87), while Cardiio Rhythm and iECG demonstrated comparable specificity (97.7%, 95% CI 97-99 vs 99.4%, 95% CI 99-100) [31]. Although Cardiio Rhythm demonstrated a lower positive predictive value (PPV) than did iECG (53.1%, 95% CI 38-67 vs 76.9%, 95% CI 56-91), both apps had high negative predictive values (NPV; 99.8%, 95% CI 99-100 vs 99.2, 95% CI 98-100) [31]. Cardiio Rhythm finger mode demonstrated 93.1% sensitivity (95% CI 86.9-97.2) and 90.9% specificity (95% CI 82.9-96.0) compared with superficial ECG, with a 92.2% PPV (95% CI 85.8-95.8) and 92.0% NPV (95% CI 94.8- 95.9) [32]. Finally, Cardiio Rhythm's facial mode effectiveness demonstrated high sensitivity (95%, 95% CI 87-98) and high specificity (96%, 95% CI 91-98) in discriminating AF compared with 12-lead ECG. The PPV and NPV of the facial mode was 92% (95 CI 84-96) and 97% (95% CI 93-99), respectively [33].

Preventicus Heartbeats is freely available for smartphones, while Preventicus Nightwatch is available only for smartwatch users. Both apps use PPG in screening for AF, and both have been validated in clinical trials.

The full version of Preventicus Heartbeats allows the user to record PPG and receive a complete report about the rhythm variability. In 2019, the Enhanced Diagnostics for Early Detection of Atrial Fibrillation–Prospective Validation (DETECT AF PRO) trial was performed to compare the efficacy of Preventicus Heartbeats in AF screening with iECG. The sensitivity and specificity of the Preventicus Heartbeats app increased with recording time from 1-3 to 5 minutes: the sensitivity was found to be 89.9% (95% CI 85.5-93.4), 91.3% (95% CI 86.5-94.7), and 91.5% (95% CI 85.9-95.4), respectively, while the specificity was found to be 99.1% (95% CI 97.5-99.8), 98.7% (95% CI 96.7-99.6), and 99.6% (95% CI 97.8-100), respectively [34].

A similar trial, Smartwatches for Detection of Atrial Fibrillation (WATCH AF), was carried out to compare the efficacy of heart rhythm monitoring by the Preventicus Nightwatch smartwatch PPG-based algorithm with that of iECG. One-minute recordings were analyzed by the Preventicus Nightwatch (available for smartwatches only) and compared with the iECG. The algorithm demonstrated 93.7% sensitivity (95% CI 89.8-96.4) and 98.2% specificity (95% CI 95.8- 99.4) in detecting AF [35,36]. Preventicus Nightwatch appear to represent a breakthrough in the monitoring of arrhythmia as it will be able to continuously analyze PPG and document AF events lasting for at least 1 minute. However, it still remains in testing [35].

Fibricheck is the only PPG-based heart rhythm monitoring app cleared both by the FDA and CE. In one study, a comparison of heart rate measurements by FibriCheck and 2 other FDA-cleared devices, Nonin oximeter and AliveCor, found a correlation of 0.834 between FibriCheck and Nonin, 0.88 between FibriCheck and AliveCor, and 0.897 between Nonin and AliveCor (no significant difference; P=.61); in addition, an R-R and peak-to-peak interval correlation of 0.993 was found between FibriCheck and wearable ECG (no significant difference; P=.92) [37]. 

FibriCheck was also included in the Real Life Digital Population Screening for Atrial Fibrillation Using only a Smartphone (DIGITAL AF II) study, including over 60,000 participants who completed the monitoring period. The study yielded a database of nearly 600,000 pieces of 1-minute PPGs [38]. Of these, 791 participants (1.3%) presented a trace typical for AF. The prevalence of AF in this population was found to be 1.68% in patients aged 40-49 years, 2.16% in those aged 50-59 years, 3.23% in those aged 60-69 years old, 5.97% in those aged 70-79 years, and 12.3% in those aged ≥80 years [38]. Unfortunately, the study has a few limitations: the traces were not compared with any other method, such as iECG or ECG, and only selected data were available. Elsewhere, FibriCheck demonstrated a sensitivity of 96% and a specificity of 91.1% compared with 12-lead ECG [39]. Its cost is not refundable from national health funds [40].

The Heart app is an integral part of iOS and watchOS. All Apple Watch series use PPG to record the heart rate, but only series 4 and above are able to record ECG (see section ECG App for Apple Watch: Series 4 or Subsequent). However, Preventicus Nightwatch will be able to use a built-in algorithm to analyze the PPG traces recorded by Apple Watch to detect AF.

Kardia Mobile (former name: AliveCor) is a clinically validated mobile device for recording 1-lead ECG and the first to be cleared by the FDA [52]. The first of 2 studies that contributed to FDA clearance of iECG was conducted by Garabelli et al [53]. The obtained ECG curve corresponds to the first (I) limb lead. The Kardia Mobile app has a built-in automatic algorithm for arrhythmia detection focused on AF.

Although some kinds of arrhythmia, like premature ventricular/supraventricular contractions or conduction abnormalities (sinus bradycardia/tachycardia, bundle branch block, or atrioventricular block) may be improperly classified as AF or even unclassified by the automatic algorithm [54,55], the app has been updated to reduce the number of unclassified traces. A study on 214 patients found the single-channel ECG to demonstrate 90.9% sensitivity (95% CI 78.3-97.5) and 93.5% specificity (95% CI 88.7-96.7) for any rhythm abnormality, and 46.4% sensitivity (95% CI 27.5-66.1) and 100% specificity (95% CI 98.0-100) for any conduction abnormality [56]. As a result, even if an automatic algorithm improperly classifies the 1-lead ECG trace as AF, it may be correctly reclassified by a clinician.

A comparison of 1-lead ECG with lead I and II of 12-lead ECG in patients taking sotalol or dofetilide found reasonable agreement between measurements of corrected QT (QTc) interval in the sinus rhythm (bias 3 ms; SD of bias 46 ms) if QTc <500 ms [57].

The efficacy of Kardia Mobile in arrhythmia detection was validated in patients with cardiovascular implantable electronic devices. A study of recordings from 251 subjects with a pacemaker (59%) or implantable cardioverter-defibrillator (41%) in paced and nonpaced states (if possible) found the readings to be adequately interpreted in 90% of paced recordings (25 of 251 recordings were “uninterpretable”) and 94.7% of nonpaced recordings (9 of 171 recordings were “uninterpretable”) [58].

Kardia Mobile is an effective tool for detecting arrhythmia or conduction abnormalities in children. It was found capable of detecting supraventricular tachycardia, AF, ectopic atrial tachycardia, atrial tachycardia, and ventricular tachycardia, and the users reported a high level of satisfaction [59]. In addition, a relationship was found between QRS dispersion and QTc intervals measured by 1-lead and 12-lead ECG in both healthy children and children with cardiac disease [60].

The QT intervals recorded by Kardia Mobile were 7 ms shorter than those from the 12-lead ECG, with only a 1.75% difference. In comparison, PQ intervals were found to be 20 ms shorter than those of conventional ECG, representing a more than 10% difference. Such a significant discrepancy between PQ intervals might lead to mimicking arrhythmias, otherwise known as pre-excitation syndrome [61].

The Kardia Mobile 6L is the first FDA-cleared 6-lead ECG. It has 3 built-in electrodes that record 6-lead ECG in channels I-III, aVR, aVL, and aVF [62]. The system uses the same app as the 1-lead Kardia Mobile. Thus far, it has not been included in clinical trials. It is expected that the new 6L will provide better-quality ECGs and greater information on ST-segment changes or axis determination than the standard AliveCor device.

Kardia Band was the first FDA-cleared medical accessory for the Apple Watch Series 1 to 3 and replaced the original band. It has a specially designed band with 2 built-in side electrodes for recording 1-lead ECG [63]. The sale of Kardia Band was terminated after the Apple Watch Series 4 was released.

Istel HR-2000 (Diagnosis SA) is a CE-certified device that has 4 built-in electrodes corresponding to 5 electrodes of a conventional ECG: the left arm, right arm, left leg, and right leg. The system records a 6-lead real-time ECG and an automatic algorithm recognizes AF. High-quality reports might be analyzed by experts if the result is ambiguous. The 6-lead ECG might serve as an event recorder, thus allowing the identification of other types of arrhythmia, like supraventricular or ventricular tachycardia, premature ventricular or supraventricular contractions, and atrioventricular blocks [64]. No specificity or sensitivity values for AF detection or the correlation status between intervals measured by the device and conventional ECG has been validated in clinical trials.

In 2018, Apple Incorporated introduced the Apple Watch Series 4, the first smartwatch to record a 1-lead ECG, corresponding with lead I from conventional ECG. Apple Watch Series 4 included 2 black crystal electrodes on the back and another electrode that serves as a Digital Crown [65]. An Apple-sponsored multicenter study with 588 patients was performed to determine the Health app's ability to generate an ECG curve corresponding to lead I from a conventional ECG and to use an algorithm classifying heart rhythms as either a sinus rhythm or AF [66]. The results were quite promising: the sensitivity for AF detection was 98.3% and the specificity was 99.6%. Consequently, the app was awarded FDA approval for Apple Watch Series 4 and above [66].

One registered clinical trial in the Cleveland Clinic has compared to the Apple Watch Series 4 and standard telemetry monitoring [67]. Recruitment has finished, but the publication of results is still pending.

Heart Beat is a free app that records heart rate using PCG [68]. PCG is a diagnostic technique that records cardiac acoustic phenomena [69] generated by interactions between the blood flow and heart chambers, valves, and great vessels [70]. A microphone must be placed on the chest to correctly measure the heart rate, with the surroundings remaining in absolute silence. The Heart Beat transforms the audio signal into heartbeat frequency [71]. Heart Beat has a few limitations: it has not been used in any clinical trials, absolute silence is needed when recording the signal, and its status for arrhythmia detection still remains unknown. Due to these limitations, the PCG app cannot be recommended for arrhythmia screening.

BeatScanner is the only app that uses SCG [72]. The app uses a very sensitive built-in accelerometer and gyroscope sensors in the smartphone to acquire microvibrations of the precordial area in reaction to heartbeats, blood flow, and respiration [73,74]. The vibrations can be studied along the superior-inferior axis (head to foot), the sinister-dexter axis (left to right), and the dorsoventral axis (back to front) [75]. The typical signal received by the gyroscope or accelerometer is called a seismocardiogram. The peaks in the seismocardiogram correspond to the opening and closing of the mitral and aortic valve [74,76]. The averaged SCG signal corresponds to ECG (Figure 4) [72,74,76]. According to Salerno and Zanetti [77], SCG might be applied to monitor the function of the left ventricle during ischemia. Paukkunen et al [78] propose that SCG may play a role in detecting atrial flutter. SCG may prove to be useful in arrhythmia detection, as the sensors are built into devices such as smartphones, and the method is noninvasive. Moreover, the sensors are cheap to develop, and the obtained signal is of high quality [75]. Unfortunately, no randomized controlled trials have compared BeatScanner with any of the methods validated for arrhythmia detection.