The impact of artificial intelligence on large vessel occlusion stroke detection and management: A systematic review meta-analysis
Bibliographic record
Abstract
Stroke remains the second leading cause of death worldwide, with many survivors facing significant disabilities. In acute stroke care, the timeless adage 'Time is brain' underscores the vital need for quick action. Innovative Artificial Intelligence (AI) technology potentially enables swift detection and management of acute ischemic strokes, revolutionizing acute stroke care towards enhanced automation. The study is registered with Prospero under CRD42024496716 and adheres to the Problem, Intervention, Comparison, and Outcomes framework (PICO). The analysis used Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We searched Embase, PubMed, DBLP, Google Scholar, IEEE Xplore, Cochrane database, IEEE, Web of Science, ArXiv, MedRxiv, and Semantic Scholar. The articles included were published between 2019 and 2023. Out of 1528 articles identified, thirty-seven met the inclusion criteria. We compared AI-augmented Large Vessel Occlusion (LVO) detection and non-AI LVO detection in various patient processing times related to emergent endovascular therapy in acute ischemic strokes. Triage Time, Door-to-Intervention Notification Time (INR), and Door-to -Arterial Puncture Time revealed an odds ratio (OR) of 0.39 (95 % CI: 0.29–0.54, p < 0.001), 0.30 (95 % CI: 0.21–0.42, p < 0.001), and 0.50 (95 % CI: 0detection 0.30–0.82, p = 0.007), respectively -- all of which had negligible heterogeneity (I^2 = 0). CT-to-Puncture-Time and Door-to-CTA-Time yielded an OR of 0.57 (95 % CI: 0.31–1.04, p = 0.065) and 0.77 (95 % CI: 0.37–1.60, p = 0.489), respectively -- both of which had negligible heterogeneity (I^2 = 0). The Last Known Well (LWK) to Time of Arrival resulted in an OR of 1.15 (95 % CI: 0.83–1.59, p = 0.409, I^2 = 0). AI stroke detection sensitivity OR of 0.91 (95 % CI: 0.88–0.95, p < 0.001) should be interpreted with potential heterogeneity in mind (I^2 = 69.3). National Institute of Health score (NIHSS) mean of 16.20 (95 % CI: 14.96–17.45, p = 0.001, I^2 = 0). Patient Transfer-Times between primary and comprehensive stroke centers generated an OR of 0.98 (95 % CI: 0.73–1.32, p = 893, I^2 = 0). Similarly, Door-in-Door-Out Time (DIDO) had an OR of 1.19 (95 % CI: 0.21–6.88, p = 0.848) and low heterogeneity (I^2 = 5.1). The results indicated significant differences across several parameters between the AI augmentation and non-AI groups. Our findings highlight how AI augments healthcare providers' ability to detect and manage strokes swiftly and accurately within acute care settings. As these technologies progress, healthcare organizations mature, and AI becomes more integrated into healthcare systems, longitudinal studies are critical in evaluating its impact on workflow efficiency, cost-effectiveness, and clinical outcomes. • Stroke remains the second leading cause of death worldwide, with many survivors facing significant disabilities. In acute stroke care, the timeless adage 'Time is brain' underscores the vital need for quick action. Innovative AI technology potentially augments healthcare providers in swiftly detecting and managing ischemic strokes. The study is a systematic review meta-analysis examining AI's impact on managing acute large vessel occlusion stroke. This study is registered with Prospero under CRD42024496716, and our findings highlight how AI augments healthcare providers' ability to detect and manage strokes swiftly and accurately within acute care settings. As these technologies progress, healthcare organizations mature, and AI becomes more integrated into healthcare systems, longitudinal studies are critical in evaluating its impact on workflow efficiency, cost-effectiveness, and clinical outcomes.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.009 | 0.005 |
| Bibliometrics | 0.002 | 0.003 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".