AI-based software predicts ease of intubation

To develop a system that could help anesthesiologists assess the most effective course of treatment for their patients, a Swiss provider of facial imaging software has received a grant from the Swiss Commission for Technology and Innovation (CTI; www.kti.admin.ch).

nViso (http://nviso.ch) will use its artificial intelligence (AI)-based facial imaging software to categorize patients who require tracheal intubation for surgeries involving general anesthesia, reducing the costs and risks for this type of procedure.

The results of a pilot project, which involved 800 patients at Lausanne University Hospital (CHUV), showed that prototype software could predict the Mallampati score for patients requiring tracheal intubation with an accuracy of 97%.

In anesthesia, the Mallampati score (http://bit.ly/bl8QE9) is used to predict the ease of intubation. It is determined by analyzing the anatomy of the oral cavity. More specifically, it is based on the visibility of the base of uvula, the arches in front of and behind the tonsils, and the soft palate.

"The leading cause of morbidity and death in general anesthesia is a result of difficulties in intubation. This stems directly from the inaccurate assessment of patients. Our results demonstrate that automated predictive analytics can offer better and more efficient ways to assess and characterize patients, compared to traditional manual and subjective approaches," says Dr. Patrick Schoettker of CHUV.

The grant will allow nViso to enhance the capabilities of its software by evaluating the features of the face, mouth, and neck, which are critical to predicting potential intubation problems. Recorded biometric data will be analyzed to generate an accurate Mallampati score.