Development and validation of a deep learning model to diagnose COVID‐19 using time‐series heart rate values before the onset of symptoms

Journal of Medical Virology, Journal Year: 2023, Volume and Issue: 95(2)

Published: Jan. 5, 2023

Abstract One of the effective ways to minimize spread COVID‐19 infection is diagnose it as early possible before onset symptoms. In addition, if can be simply diagnosed using a smartwatch, effectiveness preventing will greatly increased. this study, we aimed develop deep learning model symptoms heart rate (HR) data obtained from smartwatch. for diagnosis, proposed transformer that learns HR variability patterns in presymptom by tracking relationships sequential data. cross‐validation (CV) results unvaccinated patients, our exhibited high accuracy metrics: sensitivity 84.38%, specificity 85.25%, 84.85%, balanced 84.81%, and area under receiver operating characteristics (AUROC) 0.8778. Furthermore, validated external multiple datasets including healthy subjects, well vaccinated patients. subject group, also achieved 77.80%. patient provided similar metrics those CV: 87.23% AUROC 0.8897. dropped 66.67% 0.8072, respectively. The first finding study accurately patients HRs smartwatch second trained may provide less accurate diagnosis performance compared with last certain period time degraded performances virus continues mutate.

Language: Английский

---

Understanding the Pivotal Role of the Vagus Nerve in Health from Pandemics

Bioengineering, Journal Year: 2022, Volume and Issue: 9(8), P. 352 - 352

Published: July 29, 2022

The COVID-19 pandemic seems endless with the regular emergence of new variants. Is SARS-CoV-2 virus particularly evasive to immune system, or is it merely disrupting communication between body and brain, thus pre-empting homeostasis? Retrospective analysis AIDS pandemics, as well prion disease, emphasizes pivotal but little-known role 10th cranial nerve in health. Considering neuroimmunometabolism from point view vagus nerve, non-invasive bioengineering solutions aiming at monitoring stimulating vagal tone are subsequently discussed next optimal global preventive treatments, far beyond pandemics.

Language: Английский

---

Wearable Devices and Explainable Unsupervised Learning for COVID-19 Detection and Monitoring

Diagnostics, Journal Year: 2023, Volume and Issue: 13(19), P. 3071 - 3071

Published: Sept. 28, 2023

Despite the declining COVID-19 cases, global healthcare systems still face significant challenges due to ongoing infections, especially among fully vaccinated individuals, including adolescents and young adults (AYA). To tackle this issue, cost-effective alternatives utilizing technologies like Artificial Intelligence (AI) wearable devices have emerged for disease screening, diagnosis, monitoring. However, many AI solutions in context heavily rely on supervised learning techniques, which pose such as human labeling reliability time-consuming data annotation. In study, we propose an innovative unsupervised framework that leverages smartwatch detect monitor infections. We utilize longitudinal data, heart rate (HR), variability (HRV), physical activity measured via step count, collected through continuous monitoring of volunteers. Our goal is offer effective affordable detection employs interpretable clusters normal abnormal measures, facilitating progression detection. Additionally, enhance result interpretation by leveraging language model Davinci GPT-3 gain deeper insights into underlying patterns relationships. results demonstrate effectiveness learning, achieving a Silhouette score 0.55. Furthermore, validation using techniques yields high accuracy (0.884 ± 0.005), precision (0.80 0.112), recall (0.817 0.037). These promising findings indicate potential identifying inflammatory markers, contributing development efficient reliable methods. study shows capabilities wearables, reflecting pursuit low-cost, accessible addressing health related diseases, thereby opening new avenues scalable widely applicable solutions.

Language: Английский

---

The Role of Artificial Intelligence and Machine Learning for the Fight Against COVID-19

Springer optimization and its applications, Journal Year: 2023, Volume and Issue: unknown, P. 111 - 128

Published: Jan. 1, 2023

Language: Английский

---

Unsupervised machine learning to investigate trajectory patterns of COVID-19 symptoms and physical activity measured via the MyHeart Counts App and smart devices

npj Digital Medicine, Journal Year: 2023, Volume and Issue: 6(1)

Published: Dec. 22, 2023

Abstract Previous studies have associated COVID-19 symptoms severity with levels of physical activity. We therefore investigated longitudinal trajectories in a cohort healthcare workers (HCWs) non-hospitalised and their real-world 121 HCWs history infection who had monitored through at least two research clinic visits, via smartphone were examined. compatible provided an Apple Watch Series 4 asked to install the MyHeart Counts Study App collect symptom data multiple activity parameters. Unsupervised classification analysis identified trajectory patterns long short duration. The prevalence for persistence any was 36% fatigue loss smell being most prevalent individual (24.8% 21.5%, respectively). 8 features obtained groups high low Of these parameters only ‘distance moved walking or running’ trajectories. report long-term HCWs, method identify trends, investigate association. These highlight importance tracking from onset recovery even individuals. increasing ease collecting non-invasively wearable devices provides opportunity association other cardio-respiratory diseases.

Language: Английский

---