2025 · Heart & lung : the journal of critical care · paywalled
Can ChatGPT accurately detect atrial fibrillation using smartwatch ECG?
Avidan et al.
The finding, in our words
ChatGPT-4o identified atrial fibrillation from smartwatch ECGs with only 47.3 percent accuracy and sinus rhythm with 71.6 percent accuracy, substantially inferior to the Apple Watch algorithm's 97.3 percent and 100 percent. No patient characteristics predicted the AI's success. The findings demonstrate that human expert interpretation remains indispensable for reliable AF diagnosis from wearables, despite the theoretical appeal of large language models.
A paraphrase to the Library’s standard, never the abstract. The source is one link away and is always the authority.
A direct-to-participant, multichannel recruitment strategy enrolled 34,244 older adults into a virtual trial testing an Apple Watch based heart health programme with electrocardiogram and irregular rhythm notification features, achieving broad geographic and gender diversity but under-representing non-White ethnic groups.
Analysis of six studies revealed that decentralised clinical trial elements enhanced data completeness by enabling direct access to medical records, although this introduced a significant burden for data abstraction. The results suggest that decentralised approaches are not a universal solution but provide specific metrics that can help design fit-for-purpose trials.
In progressive multiple sclerosis, higher serum neurofilament light chain levels were associated with reduced daily step counts from wearables, while higher step counts were linked to lower sNfL. A ten percent increase in steps corresponded to a 0.015 decrease in sNfL z-score, suggesting rising sNfL may predict subsequent mobility decline.
Joseph et al., Annals of clinical and translational neurology (paywalled) · source ↗
Wearable-derived digital biomarkers have been accepted by regulatory agencies as endpoints in clinical trials for Duchenne muscular dystrophy, showing their potential to support decentralised diagnostics through patient-centric monitoring. This matters because it enables more sensitive and continuous assessment of disease progression outside clinical settings.
Ma et al., Neurology and therapy (paywalled) · source ↗
Higher plasma BDNF levels were associated with a lower prevalence of sarcopenia and better kinetic and spatiotemporal gait performance in older adults.
Zhang et al., Journal of cachexia, sarcopenia and muscle · source ↗