None listed
Conditions
Brief summary
Atrial fibrillation (AF) is the most common heart rhythm problem and a frequent reason for emergency department (ED) visits. Many patients with AF who come to the ED will spontaneously return to a normal heart rhythm on their own — a process called spontaneous cardioversion (SCV) — without needing active treatment. However, because clinicians currently cannot reliably predict which patients will revert naturally, most AF patients are admitted to hospital for monitoring and treatment, which leads to unnecessary hospitalisations. This study, called PREVAIL-AF, tests whether an artificial intelligence (AI) tool that analyses a patient's electrocardiogram (ECG, or heart tracing) can accurately predict which patients are likely to have SCV. The AI tool was previously trained using ECG data from a prior study at the same hospital. In PREVAIL-AF, the AI tool's prediction will be used to guide whether a patient is safely discharged home from the ED under a structured "wait-and-see" plan (with heart rate medications, blood thinners, and a follow-up cardiology appointment within 7 days), or admitted to hospital for standard care. This is a first-in-human pilot study aimed at testing whether this approach is feasible, safe, and accurate. The results will help refine the AI model and design a larger multi-centre study in the future. The study will be conducted at Flinders Medical Centre in Adelaide, South Australia.
Interventions
AI-ECG DECISION-SUPPORT TOOL (Intervention Arm): An artificial intelligence-enabled electrocardiogram (AI-ECG) tool — previously trained on ECG data from the REVERT-AF study — is applied to the presenting 12-lead ECG of patients with primary atrial fibrillation (AF) in the emergency department (ED). The tool analyses the ECG together with the patient's age and sex to produce a binary prediction: • "SCV predicted" — patient is likely to spontaneously cardiovert (return to sinus rhythm without i
AI-ECG DECISION-SUPPORT TOOL (Intervention Arm): An artificial intelligence-enabled electrocardiogram (AI-ECG) tool — previously trained on ECG data from the REVERT-AF study — is applied to the presenting 12-lead ECG of patients with primary atrial fibrillation (AF) in the emergency department (ED). The tool analyses the ECG together with the patient's age and sex to produce a binary prediction: • "SCV predicted" — patient is likely to spontaneously cardiovert (return to sinus rhythm without intervention) • "Persistent AF predicted" — patient is unlikely to cardiovert without active treatment Patients predicted to undergo spontaneous cardioversion (SCV) are managed under a structured "wait-and-see" discharge protocol: — Heart rate control with standardised medical therapy— Anticoagulation per Australian AF guidelines (stroke risk stratification) — Discharge home from the ED, with discharged patients provided with AliveCor Kardia devices by hospital team to document rhythm status during discharged period — Follow-up cardiology appointment within 7 days at a an outpatient AF clinic Patients predicted to remain in persistent AF receive standard hospital care as determined by the treating clinician. The AI-ECG tool is run by authorised research team members who manually export ECG files from the hospital ECG system to a secure server. The tool then analyses the file and returns the prediction result. Duration of intervention: The AI-ECG tool is applied at a single time point in the ED (at enrolment). Follow-up continues for 12 months post-enrolment.
Sponsors
Southern Adelaide Local Health Network (Flinders Medical Centre)
Study design
Allocation
Non-randomised trial
Intervention model
Single group
Primary purpose
Diagnosis
Masking
Open (masking not used)
Eligibility
Sex/Gender
All
Age
18 Years to No maximum
Healthy volunteers
No
Inclusion criteria
1. Age 18 years or older 2. Presenting to the Flinders Medical Centre Emergency Department (or affiliated SALHN ED site) 3. Primary presenting complaint of atrial fibrillation (AF) confirmed on 12-lead ECG 4. AF episode duration unknown or estimated at less than 48 hours (first-detected or paroxysmal AF with uncertain onset) 5. Haemodynamically stable (no requirement for immediate emergency cardioversion) 6. Able to provide written informed consent. 7. Able to attend a follow-up cardiology appointment within 7 days of ED discharge 8. Willing and able to comply with all study procedures and follow-up requirements
Exclusion criteria
1. Haemodynamically unstable AF requiring immediate electrical cardioversion 2. AF duration clearly greater than 48 hours 3. Permanent AF (known not to revert to sinus rhythm) 4. Prior catheter ablation for AF 5. Implanted cardiac device (pacemaker or implantable cardioverter-defibrillator [ICD]) that may interfere with ECG analysis 6. Unable to provide informed consent (e.g., significant cognitive impairment, acute delirium, or altered consciousness at time of presentation) 7. Unable to communicate in English (no interpreter available within the timeframe required for ED disposition decision) 8. Pregnant women 9. Children and young people under 18 years 10. Patients whose treating clinician determines they require immediate admission regardless of AF rhythm (e.g., for another acute medical condition)
Outcome results
None listed