False Atrial Fibrillation Alerts from Smartwatches are Associated with Decreased Perceived Physical Well-being and Confidence in Chronic Symptoms Management

Abstract

Wrist-based wearables have been FDA approved for AF detection. However, the health behavior impact of false AF alerts from wearables on older patients at high risk for AF are not known. In this work, we analyzed data from the Pulsewatch (NCT03761394) study, which randomized patients (≥ 50 years) with history of stroke or transient ischemic attack to wear a patch monitor and a smartwatch linked to a smartphone running the Pulsewatch application vs to only the cardiac patch monitor over 14 days. At baseline and 14 days, participants completed validated instruments to assess for anxiety, patient activation, perceived mental and physical health, chronic symptom management self-e cacy, and medicine adherence. We employed linear regression to examine associations between false AF alerts with change in patient-reported outcomes. Receipt of false AF alerts was related to a dose-dependent decline in self-perceived physical health and levels of disease self-management. We developed a novel convolutional denoising autoencoder (CDA) to remove motion and noise artifacts in photoplethysmography (PPG) segments to optimize AF detection, which substantially reduced the number of false alerts. A promising approach to avoid negative impact of false alerts is to employ arti cial intelligence driven algorithms to improve accuracy.

Consumer-grade wearable cardiac monitors: What they do well, and what needs work

“…However, although false alerts have been shown to reduce perceived physical well-being, the fi nancial impact of false-positive detections is not well understood. 32 As with all medical tests, the positive predictive value varies signifi cantly based on the patient population. An important tenet of Bayesian reasoning is that the posttest probability depends on the pretest probability.…”

Artificial intelligence for atrial fibrillation detection, prediction, and treatment: A systematic review of the last decade (2013–2023)

“…Among DL methods, CNNs were the most commonly used. Tran et al (2023) developed a CNN to analyze ECG data, exploring the impact of false ECG alerts on patient-reported health outcomes. Giskes et al (2023) deployed AF selfscreening stations in clinical waiting areas incorporating software to screen patient records and ECGs, identifying those potentially benefiting from treatment.…”

“…This decrease in accuracy is likely due to the increased challenge of these multi-factorial tasks. While ECG signals remain extensively studied due to their diagnostic richness (D. Han et al, 2022b;Tran et al, 2023), promising works have also leveraged diverse, complementary data sources. D. Han et al (2022b) employed a hybrid ML learning approach for assessing AF risk in ischemic stroke patients while Tran et al (2023) developed an innovative DL method for detecting false AF alerts.…”

“…While ECG signals remain extensively studied due to their diagnostic richness (D. Han et al, 2022b;Tran et al, 2023), promising works have also leveraged diverse, complementary data sources. D. Han et al (2022b) employed a hybrid ML learning approach for assessing AF risk in ischemic stroke patients while Tran et al (2023) developed an innovative DL method for detecting false AF alerts. Some studies have achieved good performance utilizing non-ECG sources such as PPG signals (Pereira et al, 2020) or integrating clinical variables from electronic health records (Giskes et al, 2023;Yue et al, 2019).…”

Study protocol for ADHERE (Applying Directly observed therapy to HydroxyurEa to Realize Effectiveness): Using small business partnerships to deliver a scalable and novel hydroxyurea adherence solution to youth with sickle cell disease

“…This protocol may also be informative for the scalability of other technology-based interventions involving large amounts of data. For example, recent studies exploring the impact of wearable technology on the identification and treatment of atrial fibrillation have identified the importance of filtering out excess data to maximize the proportion of cases requiring intervention that are escalated to the clinical team [56][57][58][59][60][61][62]. We have considered these factors here to ensure that the Scene Health team is empowered to engage with participants and assess…”