Towards a predictive model for post-stroke delirium

Hey, maybe you should look at the specific damage area in the brain. Objective observations are much more repeatable. 
http://www.naric.com/research/rehab/record.cfm?search=2&type=all&criteria=J64332&phrase=no&rec=119475
Abstract: Study assessed predisposing and precipitating risk factors for post-stroke delirium in order to create a predictive model. One hundred patients were assessed daily for delirium during the first week post stroke and risk factors were recorded. The following predisposing or precipitating risk factors for delirium were evaluated prospectively: patient characteristics, pre-stroke medical history, stroke characteristics, pharmacological factors, medical complications, and laboratory parameters. An episode of delirium was detected in 43 patients (43 percent). Higher age, metabolic disturbances, intracerebral haemorrhage and larger ischemic hemispheric strokes increase the risk of post-stroke delirium. Using multivariate logistic regression, 2 alternative predictive statistical models were developed using age, stroke type, extent of lesion, and either certain metabolic abnormalities or a more general scoring system of multiple organ dysfunction or failure. In ischemic strokes, total anterior circulation infarctions were more frequently associated with delirium (73.3 percent developed delirium) compared with the remainder of the groups combined.