Visualisation and categorisation of respiratory mechanism using self-organising maps

In post-operative patients, it is sometimes necessary to push morphine-like analgesics to their limits for pain relief. Unfortunately, this can sometimes bring a significant risk of disrupting the control of breathing, and precipitating life-threatening conditions. A possible way of monitoring patients is by studying the correlation between analgesia, airway obstruction and hypoxia. The first step towards achieving this objective is by visualising the relationship between different pairs of signals involved in respiratory mechanics. Based on previous work where self-organising maps were used for representing these relationships on a breath-by-breath basis, this paper demonstrates how it is now possible to automatically label nodes in the self-organising maps based on the classification of the signals by a clinician. In addition, the ability to visualise the transitions between categories should enable further research into the significance of transitions between the categories and the presence of possible new sub-categories.