DAS Project Grant
A new approach to predicting difficult tracheal intubation
Dr Peter Charters
Background: During any operation the anaesthetist must maintain airway patency to ensure oxygen delivery to the lungs. Tracheal intubation is the insertion of a tube via the mouth into the windpipe and is one way to maintain airway patency. A rigid blade laryngoscope is used to displace the tongue to one side of the mouth until a direct view of the vocal cords is obtained and the tube is then advanced down into the windpipe. While this is normally a simple procedure, in 1% or less of the population it is problematic because of abnormal anatomy or because the view with the laryngoscope
is poor and the tube ends up in the gullet instead of the windpipe. Unfortunately, not all patients with this problem can be recognised by clinical examination. This is important because many anaesthetics are given by people who will have little experience of these cases because they are uncommon. Our group has collaborated over many years to build a computer simulation to help understanding and management of these cases which can be quite complex.
Aims: For this current study we plan to use machine learning to investigate the most
important factors in difficult intubation and in addition to test the factors which appear to be important for their mechanical significance in our virtual biomechanical model of the human airway.
Methodology: The "Predictor" will be constructed on fairly standard machine learning
techniques but then set as an on-line facility to suggest the expected degree of difficulty or otherwise to in response to colleagues uploading anonymized patient measurements. This should encourage colleagues from the DAS to contribute to the project because they will become aware of improvements in the accuracy of the predictions as the database increases in size. These data will also be available to the project team so that the mechanical significance of the data can be tested in the computer model of the human airway as it evolves in response to this and concurrent clinical trials that will be in progress at University Hospital Aintree.
Expected outcomes:
Apart from the goal of making progress in the field of predicting difficult intubation which
has been a troublesome clinical problem for many years, the trial data should contribute
significantly to the overall verification of the accuracy of our computer simulation of the
human airway. The implications for the simulation project in the long term are far reaching.
We anticipate substantial changes in the developing and testing of new airway equipment and the potential to explore easily customizable versions of the model to test management options of individual difficult cases in computer simulation immediately before the actual clinical event.
