Delirium is a syndrome defined as an acute disturbance of both consciousness and cognition
that tends to fluctuate over time and is caused by the physiological consequences of a
medical condition. It is a common disorder in acute care settings, in internal medicine
units, in post-operative patients and the intensive care unit. Delirium is associated with
increased mortality, longer hospital stays, long-term cognitive impairment and increased
healthcare costs. The pathophysiology of delirium is multifactorial and is not completely
understood.
The prevalence of delirium increases with age and is very common in elderly hospitalized
patients. In certain departments delirium rates can reach over 40%. However, delirium is
underdiagnosed in almost two thirds of cases or misdiagnosed as depression or dementia.
Furthermore, it has been previously shown that the diagnosis of delirium is often delayed,
and that the recognition and documentation of delirium by physicians and nurses is far from
optimal. Early diagnosis of delirium may improve clinical outcome, with shortened duration of
symptoms, decreased length of admission and reduced long-term complications.
Clinical studies have demonstrated that delirium may be prevented in up to one-third of cases
by multifactored non-pharmacological interventions, yet they can be costly to implement and
require specially trained staff members. In addition, they do not usually consider
physiological parameters.
Three recent technological advances now provide opportunities for a new delirium prevention
approach. First, over the recent years vital signs monitoring with wearable sensors powered
by advanced processing algorithms has become technically feasible. This development may
provide opportunities for early detection of delirium and for detection of physiological
triggers of delirium such as dehydration, infections, and lack of sleep. Second, recent
advances in virtual dialogue systems (e.g. Amazon's Alexa or Apple's Siri) provide new and
exciting opportunities for automatic patient interaction. Devices with voice or multimodal
communication can be used by older patients with little or no experience in modern mobile
technology. Lastly, recent progress in digitized data acquisition, computing infrastructure
and algorithm development, now allow artificial intelligence and machine learning
applications to expand into areas in medicine that were previously thought to be only the
province of human experts. The combination of these three data sources can greatly improve
current prediction models and allow for earlier and more accurate delirium prediction.
An automated system which could aid with delirium detection and alert clinicians to a
possible onset of the syndrome can greatly improve treatment and outcomes for patients. The
CogMe system utilizes current technology to provide a holistic and scalable approach for
delirium prediction, detection and prevention covering both physiological and cognitive
aspects. The system uses wearables for physiological vitals monitoring and communicates with
patients by a dedicated tablet app - the CogMe Personal Assistant (PA). In this study, the
data collected by the wearables and the CogMe PA, in combination with patient data from the
EMR, will be analyzed retrospectively using machine learning techniques (CogMe Data
Analytics) to evaluate the ability of the CogMe system to predict and detect delirium.