Agitation-sedation cycling in ICU patients is characterised by oscillations between states of agitation and over-sedation. This cycling damages health and increases both length of stay and health care cost. A mathematical model that quantifies agitation is developed to achieve an agitation-feedback sedation controller with a long term goal of being used to improve their sedation management. To quantify agitation, physiological signals readily available from common patient monitoring devices are filtered and processed. Motion sensing is performed using frame-to-frame correlation of regions of interest. The resulting physiological markers are combined with motion sensing information by fuzzy inference systems to produce an agitation index. The initial goal of this objective agitation sensor is to assist the nursing staff by providing more information on agitation than currently available, in an objective and preferably preemptive manner, before it can be directly fed back to the semi-automated sedation controller. The signal-processing algorithms are first developed and validated off-line from the trials on 12 ICU patients as well as 10 healthy individuals. The agitation sensor is then statistically assessed in real-time during a second set of clinical trials. Clinical trials combined with observation increase the knowledge of agitation and permit to improve the structure of the sensor toward a reliable clinical tool.
展开▼
