To evaluate the impact of artificial intelligence (AI) imaging software on endovascular thrombectomy rates and interhospital transfer times for acute stroke patients in NHS hospitals.
Key Findings:
Endovascular thrombectomy rates doubled from 2% to 5% at evaluation sites post-AI implementation.
Non-evaluation sites showed a smaller increase from 2% to 3%.
AI use was associated with a median door-in door-out time reduction of 64 minutes.
Higher rates of intravenous thrombolysis and modestly improved functional outcomes at discharge were observed with AI support.
No significant association was found between AI use and in-hospital mortality.
Interpretation:
AI imaging support significantly improved treatment rates and efficiency in acute stroke care, particularly in primary stroke centers with limited specialist resources.
Limitations:
Observational design limits causal inference.
Residual confounding may exist despite adjustments.
Patients reviewed with AI had more severe strokes but better premorbid function.
Limited follow-up data restricted long-term outcome evaluations.
Reasons for nonuse of AI were not captured.
Conclusion:
The implementation of AI imaging software in stroke care settings is associated with improved treatment rates and reduced transfer times, highlighting its potential to enhance acute stroke management.
