Evaluation of the performance of a machine learning based atrial fibrillation screening algorithm using an oscillometric blood pressure monitor

Blood pressure monitors (BPMs) with atrial fibrillation (AFib) detection function can be used to detect AFib early. However, conventional algorithms require multiple BP measurements. Here, the feasibility of a machine-learning-based approach for AFib detection through single BP measurement was evaluated. First, a custom AdaBoost-based software, which analyzes the pulse-to-pulse interval (PPI) pattern and classifies it based on AFib detection, was created. Then, its classification performance was validated. For the validation study, PPI and standard 12-lead electrocardiogram (ECG) datasets were collected from 79 and 92 Japanese participants with and without AFib, respectively. PPI data were obtained using two different BPMs. All ECG results were interpreted by cardiologists. The custom software output for the PPI dataset and ECG interpreted results was compared, and the sensitivity and specificity were calculated. A sensitivity and specificity for PPI from main device were 97.5% (95% confidence interval [CI] 91.2-99.3%) and 98.9 (95% CI 94.1-99.8), respectively. No significant differences in sensitivity and specificity were observed in the subgroup analysis between different devices, age groups, and arm size groups. These results reflect the high accuracy and robustness of this AFib algorithm using a single BP measurement and supports its use for widespread AFib screening.

Simple risk scoring using sinus rhythm electrocardiograms predicts the incidence of atrial fibrillation in the general population

Atrial fibrillation (AF) is an arrhythmic disease. Prediction of AF development in healthy individuals is important before serious complications occur. We aimed to develop a risk prediction score for future AF using participants' data, including electrocardiogram (ECG) measurements and information such as age and sex. We included 88,907 Japanese participants, aged 30-69 years, who were randomly assigned to derivation and validation cohorts in a ratio of 1:1. We performed multivariate logistic regression analysis and obtained the standardised beta coefficient of relevant factors and assigned scores to them. We created a score based on prognostic factors for AF to predict its occurrence after five years and applied it to validation cohorts to assess its reproducibility. The risk score ranged from 0 to 17, consisting of age, sex, PR prolongation, QT corrected for heart rate prolongation, left ventricular hypertrophy, premature atrial contraction, and left axis deviation. The area under the curve was 0.75 for the derivation cohort and 0.73 for the validation cohort. The incidence of new-onset AF reached over 2% at 10 points of the risk score in both cohorts. Thus, in this study, we showed the possibility of predicting new-onset AF using ECG findings and simple information.

Regular pulse checks for patients with non-cardioembolic stroke in rehabilitation hospitals to improve recognition and detection of atrial fibrillation (the ESCORT study): protocol for a prospective multicenter observational study

Background

Cryptogenic stroke (CS) are heterogeneous in origin; however, most CS are embolic mechanism. Paroxysmal atrial fibrillation (AF) is suspected to be a major type of CS that leads to severe cerebral infarction without anticoagulant use. Therefore, the identification of AF is vital in patients with CS. However, patients are often unaware of AF because they have no symptoms, and AF may not be detected on an electrocardiogram (ECG) or Holter ECG on admission. After patients with stroke are treated in the acute phase, they are promptly transferred to a rehabilitation hospital for functional recovery. Once the patient is transferred to a hospital, a few attempts are made to detect AF. In addition, rehabilitation therapists are considered to have insufficient awareness of the possibility of undiagnosed AF.

Objective

This study aimed to increase the understanding of the importance of AF detection in patients with ischemic stroke among therapists in rehabilitation hospitals and to investigate whether regular pulse screening can aid in the detection of AF. If AF was detected, we determined the rate and timing of AF detection and identified the patient characteristics.

Methods

This multicenter prospective observational study aimed to detect AF in patients with non-cardiac stroke at rehabilitation hospitals. Therapists performed pulse checks before, during, and after rehabilitation. If arrhythmia or tachycardia was detected, an ECG was performed, and the physician checked for AF. If the patient complained of chest symptoms, electrocardiography (ECG) was performed to check for AF. We investigated the characteristics, laboratory data, cognitive status, complications, such as stroke recurrence, and functional outcomes of patients with AF.

Results

The study is in the enrollment phase. Recruitment began in September 2022 and will end in August 2023. Patients have provided written informed consent. The main results have been submitted for publication in your journal.

Conclusion

The findings of this study will help identify patients with AF in rehabilitation hospitals and improve awareness among therapists.

Treatment Outcomes of 94 Cases of Pipeline Embolization Device in a Single Center: Predictive Factors of Incomplete Aneurysm Occlusion

Objective

This study aimed to report the outcome of an endovascular treatment with a pipeline embolization device (PED) at a single center. We also examined the predictive factors for an incomplete occlusion after the PED placement.

Methods

The subjects were 94 patients with 109 aneurysms who underwent the PED placement at our single center from June 2015 to September 2022. As treatment outcomes, we investigated the PED placement success rate, perioperative morbidity and mortality, postoperative cranial nerve improvement rate, and the classification of angiographic result at 6 months after the PED placement. Furthermore, the predictors of an incomplete occlusion were investigated in detail.

Results

One hundred nine aneurysms locations were: C1 (9), C2 (30), C3 (15), C4 (53), and C5 (2) in the internal carotid artery segments. Perioperative morbidity, including the asymptomatic ones, occurred in 10 cases (10.6%). Among these 10 cases, the modified Rankin Scale (mRS) improved to preoperative mRS after 90 days in 9 cases except 1 case. On the other hand, no perioperative mortality was observed. The postoperative cranial nerve improvement rate was 84.4%, and 61.7% of patients had a complete occlusion in the follow-up angiography, 6 months after the PED placement. Predictive factors for an incomplete occlusion after the PED placement were the elderly aged 70 years or older (P-value = 0.0214), the elderly aged 75 years or older (P-value = 0.0009), and the use of anticoagulants (P-value = 0.0388) in an univariate analysis. Further, the multivariate analysis revealed that the elderly aged 75 years or older was a predictive factor of an incomplete occlusion in this study.

Conclusion

We summarized the outcomes of the PED treatment at our single center. In this study, the elderly aged 75 years or older was a predictive factor of an incomplete occlusion after the PED placement.

Global, regional, and national burden of heart failure associated with atrial fibrillation

BACKGROUND

Heart failure is a leading cause of mortality and morbidity worldwide, and Atrial fibrillation (AF) is among many modifiable risk factors for heart failure. No estimates are available on the magnitude of the burden of heart failure associated with AF, and this study estimated the global, regional, and national burdens associated with AF.

METHODS

We used the comparative risk assessment method to estimate the disease burden in terms of prevalence and years lived with disability (YLD). The population-attributable fraction for heart failure and AF was calculated from prevalence estimates of AF and the recalculated relative risks of heart failure associated with AF from a systematic review summarising the longitudinal association between AF and outcomes. The burden of heart failure was retrieved from the Global Burden of Disease database.

RESULTS

Globally, 2.6% (95% uncertainty interval 1.3 to 4.7%) of the burden of heart failure is associated with AF. This was 1.5 (95% UI 0.6 to 3.2) million people in 2019, a 49.8% increase from 1990. The highest prevalence was from South-East Asia, East Asia and Oceania. The highest YLD was estimated for Central Europe, Eastern Europe and Central Asia. High-income countries showed a sharp decline in the age standardised prevalence and YLD rates from 1990 to 2019.

CONCLUSION

The burden of heart failure associated with AF has increased substantially over the past two decades despite the advances in AF management. However, falling prevalence and YLD rates of heart failure associated with AF in high-income countries over time indicate that reducing this burden is possible.

Rationale and design of the NEO-NORMAL-AF study examination of the usefulness of implantable loop recorder for arrhythmia detection including atrial fibrillation in heart failure with non-reduced ejection fraction cases: a pilot study

INTRODUCTION

The incidence of arrhythmia in heart failure with non-reduced ejection fraction (HFnon-rEF) in patients who have a history of hospitalisation is unclear. The aim of this study is to investigate the usefulness of an implantable loop recorder (ILR) for arrhythmia detection including atrial fibrillation (AF) in HFnon-rEF patients after discharge.

METHODS AND ANALYSIS

This is a multicentre single arm study to evaluate the usefulness of ILR for detecting arrhythmia. The eligible patients are HFnon-rEF patients (left ventricular ejection fraction ≥40%) aged ≥20 years with a history of hospitalisation. The ILR will be implanted for qualified patients, and ECGs will be monitored and recorded for 1 year to check for arrhythmias. The primary endpoint is new-onset 6 min or more persistent AF detected by ILR. Secondary endpoints are 30 s or more persistent supraventricular tachycardia and ventricular tachycardia, 3 s or more persistent pause, bradycardia with 40 beats per minutes or lower heart rate, AF burden, all-cause death, cardiovascular death, hospital readmission due to exacerbation of HF, acute coronary syndrome, ischaemic or haemorrhagic stroke, non-pharmacological therapy such as pacemaker implantation and ablation.

CONCLUSIONS

This study is expected to provide valuable findings regarding arrhythmia in HFnon-rEF patients, and elucidate a potential new therapeutic approach for HFnon-rEF.

TRIAL REGISTRATION NUMBER

This trial has been registered in the Japan Registry of Clinical Trials (jRCT) (Trial Registration: jRCTs052210060).

Diagnosis of Atrial Fibrillation Using Machine Learning With Wearable Devices After Cardiac Surgery: Algorithm Development Study

Background

Some attempts have been made to detect atrial fibrillation (AF) with a wearable device equipped with photoelectric volumetric pulse wave technology, and it is expected to be applied under real clinical conditions.

Objective

This study is the second part of a 2-phase study aimed at developing a method for immediate detection of paroxysmal AF, using a wearable device with built-in photoplethysmography (PPG). The objective of this study is to develop an algorithm to immediately diagnose AF by an Apple Watch equipped with a PPG sensor that is worn by patients undergoing cardiac surgery and to use machine learning on the pulse data output from the device.

Methods

A total of 80 patients who underwent cardiac surgery at a single institution between June 2020 and March 2021 were monitored for postoperative AF, using a telemetry-monitored electrocardiogram (ECG) and an Apple Watch. AF was diagnosed by qualified physicians from telemetry-monitored ECGs and 12-lead ECGs; a diagnostic algorithm was developed using machine learning on the pulse rate data output from the Apple Watch.

Results

One of the 80 patients was excluded from the analysis due to redness caused by wearing the Apple Watch. Of 79 patients, 27 (34.2%) developed AF, and 199 events of AF including brief AF were observed. Of them, 18 events of AF lasting longer than 1 hour were observed, and cross-correlation analysis showed that pulse rate measured by Apple Watch was strongly correlated (cross-correlation functions [CCF]: 0.6-0.8) with 8 events and very strongly correlated (CCF>0.8) with 3 events. The diagnostic accuracy by machine learning was 0.9416 (sensitivity 0.909 and specificity 0.838 at the point closest to the top left) for the area under the receiver operating characteristic curve.

Conclusions

We were able to safely monitor pulse rate in patients who wore an Apple Watch after cardiac surgery. Although the pulse rate measured by the PPG sensor does not follow the heart rate recorded by telemetry-monitored ECGs in some parts, which may reduce the accuracy of AF diagnosis by machine learning, we have shown the possibility of clinical application of using only the pulse rate collected by the PPG sensor for the early detection of AF.

Increase in Heart Rate-dependent Left Atrial Pressure is Associated with Symptoms in Patients with Paroxysmal Atrial Fibrillation: Conclusion

BACKGROUND

It remains unclear why some patients with the same heart rate (HR) during an atrial fibrillation (AF) have subjective symptoms, whereas others do not. We assessed the hypothesis that different responses of arterial and left atrial blood pressures to rapid stimulation may be associated with the symptoms of AF.

METHODS

A total of 110 patients who underwent catheter ablation for paroxysmal AF were retrospectively studied. Asymptomatic AF was defined as a European Heart Rhythm Association (EHRA) score of Ⅰ for AF-related symptoms. The left atrial pressure (LAP) was measured during sinus rhythm (SR), in 10 pacing per minute (ppm) increments from 100 ppm to the Wenckebach block rate in high right atrial pacing.

RESULTS

Asymptomatic AF was observed in 19/110 patients (17%). Patients with symptomatic AF showed higher E/e' ratio and gradual LAP increase that was dependent on the pacing rate. Patients with asymptomatic AF had decreased LAP at 100 ppm compared that at SR, and thereafter, LAP gradually increased depending on the pacing rate. The rate of LAP change compared to that at SR was significantly lower in patients with asymptomatic AF than that in patients with symptomatic AF. The rate of LAP change was independently associated with AF symptoms. This article is protected by copyright. All rights reserved.

Assessment of a new algorithm to detect atrial fibrillation in home blood pressure monitoring device among healthy adults and patients with atrial fibrillation

The authors evaluated the diagnostic accuracy of a new algorithm for detecting atrial fibrillation (AF) using a home blood pressure (BP) monitor. Three serial BP values were measured in 205 subjects with sinus rhythm and 75 subjects with AF confirmed by electrocardiogram. Irregular pulse peak (IPP) 15 was defined as follows: |interval of pulse peak ‐ the average of the interval of the pulse peak| ≥ the average of the interval of the pulse peak × 15%. Irregular heartbeat (IHB) was defined as follows: beats of IPP ≥ total pulse × 20%. The sensitivities of IPP15 for diagnosing AF defined as two or three IHBs of three readings were 1.0 and 0.99, and the corresponding specificities were 0.97 and 0.99, respectively. The algorithm using two or more IHBs of three readings in the setting of IPP15 had the highest diagnostic accuracy for AF.

Arterial Stiffness Is Associated With Increased Symptom Burden in Patients With Atrial Fibrillation

BACKGROUND

Increased arterial stiffness (AS) has been described as a predictor of atrial fibrillation (AF). This study was performed to assess whether increased AS leads to a higher symptom burden in patients with AF.

METHODS

One hundred sixty-two consecutive patients (104 male, 58 female) with diagnosed AF (paroxysmal or persistent) were enrolled. Symptoms most likely attributable to AF were quantified according to the Canadian Cardiovascular Society Severity of Atrial Fibrillation (SAF) scale. AS indices (aortic distensibility, cyclic circumferential strain, and aortic compliance) were characterized using transoesophageal echocardiography.

RESULTS

The cohort was divided into asymptomatic to oligosymptomatic (SAF scale 0-1, n = 78 [48.1%]) and symptomatic (SAF scale ≥ 2, n = 84 [51.9%]) patients. Symptomatic patients tended to be younger (median, 75 [interquartile range (IQR) 67-80] vs 71 [65-79]; P = 0.047) and were more likely to be female (22 [28.2%] vs 36 [42.9%]; P = 0.052). Hypertension was more frequent in symptomatic patients. Aortic compliance indices each were reduced in symptomatic patients, most pronounced for aortic compliance (median, 0.05 [IQR 0.03-0.06] vs 0.04 [0.03-0.05] cm/mm Hg; P = 0.01) followed by cyclic circumferential strain (median, 0.09 [IQR 0.07-0.11] vs 0.07 [0.04-0.10]; P = 0.02) and aortic distensibility (10^-3 mm Hg^-1, median, 1.74 [IQR 1.34-2.24] vs 1.54 [1.12-2.08]; P = 0.03). Multivariable analysis revealed aortic compliance as an independent predictor for symptoms in patients with AF with an odds ratio of 2.6 (95% confidence interval, 1.2-3.4; P = 0.003).

CONCLUSIONS

AS contributes to a high symptom burden in patients with AF, emphasizing the prognostic role of AS in the early detection and prevention in patients with AF.

Use of a Smart Watch for Early Detection of Paroxysmal Atrial Fibrillation: Validation Study

Background

Wearable devices with photoplethysmography (PPG) technology can be useful for detecting paroxysmal atrial fibrillation (AF), which often goes uncaptured despite being a leading cause of stroke.

Objective

This study is the first part of a 2-phase study that aimed at developing a method for immediate detection of paroxysmal AF using PPG-integrated wearable devices. In this study, the diagnostic performance of 2 major smart watches, Apple Watch Series 3 and Fitbit (FBT) Charge HR Wireless Activity Wristband, each equipped with a PPG sensor, was compared, and the pulse rate data outputted from those devices were analyzed for precision and accuracy in reference to the heart rate data from electrocardiography (ECG) during AF.

Methods

A total of 40 subjects from patients who underwent cardiac surgery at a single center between September 2017 and March 2018 were monitored for postoperative AF using telemetric ECG and PPG devices. AF was diagnosed using a 12-lead ECG by qualified physicians. Each subject was given a pair of smart watches, Apple Watch and FBT, for simultaneous pulse rate monitoring. The heart rate of all subjects was also recorded on the telemetry system. Time series pulse rate trends and heart rate trends were created and analyzed for trend pattern similarities. Those trend data were then used to determine the accuracy of PPG-based pulse rate measurements in reference to ECG-based heart rate measurements during AF.

Results

Of the 20 AF events in group FBT, 6 (30%) showed a moderate or higher correlation (cross-correlation function>0.40) between pulse rate trend patterns and heart rate trend patterns. Of the 16 AF events in group Apple Watch (workout [W] mode), 12 (75%) showed a moderate or higher correlation between the 2 trend patterns. Linear regression analyses also showed a significant correlation between the pulse rates and the heart rates during AF in the subjects with Apple Watch. This correlation was not observed with FBT. The regression formula for Apple Watch W mode and FBT was X=14.203 + 0.841Y and X=58.225 + 0.228Y, respectively (where X denotes the mean of all average pulse rates during AF and Y denotes the mean of all corresponding average heart rates during AF), and the coefficient of determination (R²) was 0.685 and 0.057, respectively (P<.001 and .29, respectively).

Conclusions

In this validation study, the detection precision of AF and measurement accuracy during AF were both better with Apple Watch W mode than with FBT.

Clinical implication of disturbed left atrial phasic functions in the heterogeneous population associated with hypertension or atrial fibrillation

BACKGROUND

To evaluate left atrial (LA) phasic functions in patients with hypertension and/or paroxysmal atrial fibrillation (PAF) and its clinical significance.

METHODS

LA strain was studied in 77 patients (25 hypertension, 24 lone AF, and 28 with both hypertension and PAF) and 28 controls using two-dimensional speckle-tracking echocardiography (2D STE). The following indexes during atrial reservoir, conduit and pump phase were analyzed respectively: (1) peak atrial longitudinal strain (PALS) and strain rate (PALSR), (2) the standard deviation of time to PALS and PALSR of all LA segments (TpS-SD% and TpSR-SD%).

RESULTS

Compared with controls, PALS_res, PALS_cond and PALSR_cond were significantly reduced in patients with isolated hypertension (all P < 0.01) but no significant differences were observed in PALS_pump, PALSR_pump and TpS_pump-SD% between them (all P > 0.05). PALS_pump, PALSR_pump and PALSR_res were significantly lower in patients with both hypertension and PAF than in those with isolated hypertension (all P < 0.05). PALS and PALSR were significantly decreased, and TpS-SD% was significantly increased during each phase in lone AF patients than in controls (all P < 0.05), and PALSR_pump was further depressed in patients with both hypertension and PAF (P = 0.029). PALSR_cond ≤ 1.475 s^- 1 combined with TpS_pump-SD% ≥ 3.25% (sensitivity, 85%; specificity, 71%; AUC = 0.845, P < 0.001) could distinguish lone AF from healthy subjects effectively, while in hypertensive patients, PALS_pump ≤ 14.2% was found to be an independent differentiator for occurrence of AF or not with sensitivity of 81% and specificity of 84% (AUC = 0.838, P < 0.001). LAVI≥29.3 mL/m² was an independent characteristic for reflecting different LA remodeling in lone AF or hypertension with AF.

CONCLUSIONS

The impairment of LA phasic functions was varied in patients with hypertension and/or AF. The disturbed LA phasic functions were proved to have independent abilities of differential diagnosis in this heterogeneous population associated with hypertension or AF.