https://pubmed.ncbi.nlm.nih.gov/38093485
A new study published in the journal Anaesthesia has investigated the use of artificial intelligence (AI) in predicting difficult videolaryngoscopy, which is a challenging and critical aspect of airway management during anesthesia. The study developed a deep learning-based facial analysis model that uses a neural network to identify patients at risk of difficult videolaryngoscopy.
The researchers collected data from 5849 patients and used several predictor variables, including medical history, bedside examination, and seven facial images. They used ResNet-18, a convolutional neural network, to recognize and extract features from the facial images. They then used different machine learning algorithms to develop predictive models.
The facial model using the Light Gradient Boosting Machine algorithm showed the highest predictive performance, with an area under the curve of 0.779 (95% CI: 0.733-0.825), a sensitivity of 0.757 (95% CI: 0.650-0.845), and a specificity of 0.721 (95% CI: 0.626-0.794). The facial model outperformed other traditional methods, such as bedside examination and multivariate scores (El-Ganzouri and Wilson), in predicting difficult videolaryngoscopy.
The study suggests that AI-based facial analysis is a feasible and accurate technique for predicting difficult videolaryngoscopy, and it has the potential to improve airway management during anesthesia. The researchers believe that their model can help anesthesiologists identify high-risk patients and take appropriate measures to ensure safe airway management.
Overall, this study highlights the potential of AI in improving clinical outcomes and patient safety in anesthesia. Further studies are needed to validate the model’s performance in different populations and settings.
