Dies ist eine Übersichtsseite mit Metadaten zu dieser wissenschaftlichen Arbeit. Der vollständige Artikel ist beim Verlag verfügbar.
EP010 Can assistive artificial intelligence facilitate ultrasound image acquisition in the absence of formalized USGRA training?
0
Zitationen
7
Autoren
2024
Jahr
Abstract
<h3></h3> <b>Please confirm that an ethics committee approval has been applied for or granted:</b> Not relevant <h3>Background and Aims</h3> Formalised ultrasound guided regional anaesthesia (USGRA) training is resource intensive and often difficult to access. Assistive artificial intelligence (AI) is an emerging technology with potential to enhance training and provision of USGRA. We aim to evaluate if ScanNAV(TM) (Intelligent Ultrasound Limited) can enhance USGRA image acquisition of a ‘Plan-A Block’ for the non-expert in the absence of formalised training. <h3>Methods</h3> 18 anaesthetists performed sonoanatomy on live models for two pre-selected Plan-A blocks, one with and one without prior formal training. ScanNAV(TM) was used in the latter alongside ScanNAV(TM) tutorial videos and RA-UK infographic material for reference. 2 expert assessors made objective assessments for each using a protocolised data collection tool. <h3>Results</h3> 15/18 (83.3%) participants successfully acquired appropriate ultrasound images for a Plan-A Block using ScanNAV(TM) and reference materials with no formal prior training. Sonoanatomy scans were performed faster on average in block procedures that had received prior formal USGRA training. The adductor canal block had an average procedure time of 23.52seconds vs 179.17seconds (t-test value 2.74; p-value 0.0168). Where formal training had taken place, participants scored higher in identifying key structures (ASRA-ESRA Delphi consensus) across all Plan-A blocks, as well as accuracy grade of image acquisition and needle path safety. <h3>Conclusions</h3> Assistive AI, e.g. ScanNAV(TM) may facilitate image acquisition and identification of key sonoanatomical stuctures in the absence of formalised training. This technology should be used as an adjunct, not a replacement, for formalised training as objective assessment in speed, accuracy and safety were seen to be superior in this subgroup.
Ähnliche Arbeiten
Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI
2019 · 8.260 Zit.
Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead
2019 · 8.116 Zit.
High-performance medicine: the convergence of human and artificial intelligence
2018 · 7.493 Zit.
Proceedings of the 19th International Joint Conference on Artificial Intelligence
2005 · 5.776 Zit.
Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI)
2018 · 5.438 Zit.