Why the study?
Objective assessment of surgeons procedural skills and intraoperative decision making remains challenging, and surgical workflow analysis typically relies on video rather than speech.
Does speech-based classification using NLP accurately identify surgical phases during laparoscopic cholecystectomy?
Population
Audio recordings and transcriptions from educational operating rooms during laparoscopic cholecystectomy
Comparison
Four feature extraction techniques and four machine learning models
Design
Model development and validation study
Key result
Speech-based classification of laparoscopic cholecystectomy phases using an SVM coupled to a hidden-Markov model with Word2Vec features achieved an 82.95% average accuracy.
Authors
Loading...
May support audio-based surgical workflow analysis; leaves open prospective validation before training use.
Does speech-based classification using NLP accurately identify surgical phases during laparoscopic cholecystectomy?
Speech-based classification using NLP can effectively identify surgical phases during laparoscopic cholecystectomy, laying the foundation for audio-based surgical workflow analysis in training.
Guzmán-García et al. (2021) studied Laparoscopic cholecystectomy. Speech-based surgical phase recognition using NLP (SVM + HMM with Word2Vec) vs. Other machine learning models and feature extraction techniques was evaluated on Phase recognition accuracy. Speech-based classification of laparoscopic cholecystectomy phases using an SVM coupled to a hidden-Markov model with Word2Vec features achieved an 82.95% average accuracy.