A clinical dilemma: the longer the higher the risk?—Authors’ reply
Bibliographic record
Abstract
We thank Babayiğit et al.1 for their interest in our work.2 We simulated 14-day Holter monitors using the heart rhythm profiles of patients with a newly implanted cardiac rhythm device who were enrolled in Asymptomatic Atrial Fibrillation and Stroke Evaluation in Pacemaker Patients and the Atrial Fibrillation Reduction Atrial Pacing Trial (ASSERT).2,3 We estimated that 3.1% of the study population would have more than 6 min of atrial fibrillation detected by a 14-day Holter and that these patients carried an annual risk of stroke of approximately 2.2%. Among ASSERT participants, approximately 95% had a pacemaker, and the remaining 5% had an implantable cardioverter defibrillator. The proportion of study participants with a history of heart failure was 14.5%. ASSERT participants were enrolled between 2004 and 2009, when semi-quantitative echocardiography was standard. Accordingly, the structural and functional heart characteristics of this population were not characterized to the degree they are in today’s practice. Moreover, medical therapy was different. The proportion of patients who were taking beta-blockers at study entry was 36.6%; many of our other modern medical therapies were not yet standard. The ultimate goal of atrial fibrillation screening is to prevent strokes in at-risk individuals.4 We live in an era where we have myriad tools to assist us in this pursuit. Our arsenal for detecting atrial fibrillation ranges from the timeless 12-lead electrocardiogram to implantable devices to direct-to-consumer technologies that can generate a surface electrocardiogram.5 There are emerging techniques and assays for quantifying individuals’ atrial fibrillation risk, including biomarkers, functional imaging, polygenic risk scores, and artificial intelligence. The challenge is to harness these technologies and deliver the right strategy for the right patients. Observational studies generate hypotheses that can be tested in appropriately designed randomized clinical trials. Our article generates the hypothesis that patients over the age of 65 with hypertension who have an atrial fibrillation burden of greater than 6 min on a 14-day screening Holter monitor will benefit from oral anticoagulation for prevention of ischaemic stroke. Randomized clinical trials can only test one hypothesis at a time. Currently, there are more than 15 published or ongoing randomized trials that are evaluating atrial fibrillation screening. These trials differ widely in the health care settings in which they are conducted, the participants they enrol, and the modalities that are employed to detect and define atrial fibrillation. Two large, international, and co-operating consortia [AF-SCREEN (https://www.afscreen.org/) and AFFECT-EU (http://affect-eu.eu/)] have been working to systematically record and co-ordinate efforts in atrial fibrillation screening research. Among the ongoing initiatives is a prospectively planned individual participant data meta-analysis. This project is designed to offer the power to evaluate stroke as an outcome and to explore the population, screening modality, and health system factors that contribute to successful screening programmes. The primary results of ASSERT were published 10 years ago, when our understanding of atrial fibrillation was simpler and our technologies vastly more limited. However, the uniqueness of a large data set of continuously monitored patients with minimal (<2%) use of oral anticoagulation continues to advance theories that can be tested in the modern research arena. Conflict of interest: None declared.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.006 | 0.071 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.005 | 0.005 |
| Scholarly communication | 0.008 | 0.006 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.090 | 0.070 |
| Insufficient payload (model declined to judge) | 0.008 | 0.007 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".