A Machine Learning Approach for Investigating Delirium as a Multifactorial Syndrome

Abstract

Delirium is a psycho-organic syndrome common in hospitalized patients, especially the elderly, and is associated with poor clinical outcomes. This study aims to identify the predictors that are mostly associated with the risk of delirium episodes using a machine learning technique (MLT). A random forest (RF) algorithm was used to evaluate the association between the subject's characteristics and the 4AT (the 4 A's test) score screening tool for delirium. RF algorithm was implemented using information based on demographic characteristics, comorbidities, drugs and procedures. Of the 78 patients enrolled in the study, 49 (63%) were at risk for delirium, 32 (41%) had at least one episode of delirium during the hospitalization (38% in orthopedics and 31% both in internal medicine and in the geriatric ward). The model explained 75.8% of the variability of the 4AT score with a root mean squared error of 3.29. Higher age, the presence of dementia, physical restraint, diabetes and a lower degree are the variables associated with an increase of the 4AT score. Random forest is a valid method for investigating the patients' characteristics associated with delirium onset also in small case-series. The use of this model may allow for early detection of delirium onset to plan the proper adjustment in healthcare assistance.

Keywords: aging; delirium; machine learning technique; nursing; random forest.

Figure 1

Proportions of patients reporting delirium and risk for delirium during the first five days of hospitalization. The delirium state indicates a 4AT score higher than 4; the delirium or cognitive impairment (CI) state indicates a 4AT score higher than 1.

Figure 2

Variables with top importance selected by the RF algorithm according to the permutation approach. The total score measures the predictive impact of the variables, i.e., the relative decrease of the algorithm’s predictive error produced by a variable.

Figure 3

Effect of age on the 4AT delirium score. Expected delirium score estimated with random forest has been reported on the y axis according to ages with 95% confidence bounds.

Figure 4

Effect of the presence of dementia, gender, physical restraint, educational level, diabetes, ward, antibiotics, and previous admissions on the 4AT delirium score. The vertical axis displays the ensemble expected predicted delirium score.