Photo Credit: sudok1
The following is a summary of “Artificial Intelligence Guided Bronchoscopy is Superior to Human Expert Instruction for the Performance of Critical-Care Physicians: A Randomized Controlled Trial,” published in the March 2025 issue of Critical Care Medicine by Agbontaen et al.
Researchers conducted a retrospective study to compare bronchoscopy training using an artificial intelligence (AI) navigational system with bedside expert instruction in improving performance.
They performed a nonblinded, parallel-group randomized controlled trial (RCT) in a simulated setting at an academic university hospital. Critical-care physicians were invited to participate, 40 individuals received 30 minutes of bronchoscopy training, either with AI guidance alone (artificial intelligence group [AIG]) or with expert tutor feedback (expert tutor group [ETG]). All participants completed a final full-navigation bronchoscopy performance test and a cognitive load assessment using the NASA Task Load Index. (Nonblinded, parallel-group RCT, AI, bronchoscopy, academic university hospital, AIG, ETG, NASA Task Load Index).
The results showed that bronchoscopic performance efficiency, measured by mean intersegmental time (MIT), was significantly improved in the AIG compared to the ETG (median difference: –7.9 seconds, P =0.027). Procedure time (PT) was reduced by 77 seconds (P =0.022), and the number of segments revisited (SR) decreased by 7 segments (P =0.019) in the AIG. Cohen’s classification indicated moderate effect sizes for MIT (0.35), PT (0.36), and SR (0.37). No significant differences were observed between groups for other final test measures. (RCT, AIG, ETG, bronchoscopy, MIT, PT, structured progress [SP], SR, diagnostic completeness [DC], Cohen’s classification).
Investigators concluded that AI-assisted training improved bronchoscopy performance in critical-care physicians compared to expert tutors.
Source: journals.lww.com/ccmjournal/fulltext/9900/artificial_intelligence_guided_bronchoscopy_is.493.aspx
Create Post
Twitter/X Preview
Logout