Jan Claassen - US grants

Patients increasingly survive critical care, but for many the experience is psychologically tormenting in part because they lose all control over their environment. Patients with neurological injury in particular suffer from this major loss of autonomy which impacts their long term recovery. Brain-computer interface (BCI) technology transforms physiological changes associated with patient thoughts into actionable outputs. The technology is rapidly advancing and well established outside of the critical care setting. The goal of this project is to determine whether conscious and unconscious appearing ICU patients are able to communicate their basic needs, such as pain, hunger, and thirst, to health care providers using the BCI technology. We will apply a bedside BCI system in the ICU using a machine-learning algorithm that analyzes changes in routine EEG in response to standardized questions presented to the patient by headphones. The BCI system will provide immediate closed- loop feedback (auditory and/or visual) to patients about their performance. To achieve our goals, we will first test if conscious ICU patients are able to express basic needs, such as pain, to health care providers using the BCI technology. Patients will trigger the BCI system to then express their needs in response to specific questions (e.g., ?Do you have pain??). Secondly, we will determine whether basic communication is possible for patients with cognitive motor dissociation (i.e., patients who appear unconscious, but follow commands using EEG motor imagery paradigms). We will test if these patients are able to use the BCI system to respond to simple questions (i.e., ?Activate the alarm?). Finally, we will assess the patient's experience using BCI technology as part of this study, as well as acceptance of using BCI technology in future clinical trials from patients, families, and health care providers involved in the study. The long-term goal of this proposal is to prepare a large clinical trial to test the benefits of BCI-assisted communication in brain-injured ICU patients.

Unconsciousness is common after acute brain injury such as a brain hemorrhage, and recovery is poorly understood. This lack of knowledge is a key impediment to the development of novel strategies to improve outcomes, and is one of the main reasons that prognostication of recovery of consciousness and functional outcomes is inaccurate. One fifth of clinically unconscious patients with acute brain injury are able to follow commands using a simple, bedside EEG motor imagery test that directly measures brain activity associated with the attempt to move. This state is called cognitive motor dissociation (CMD). Pilot data indicates that CMD patients are more likely to clinically recover consciousness and have better long-term functional outcomes than non-CMD patients. To integrate these findings into clinical practice we need to better understand the trajectory of CMD. This will only be possible in a tightly-controlled study with a homogenous patient cohort that is well characterized early after the injury and captures long-term outcomes. The over-arching hypothesis of this application is that once confounders are accounted for, recovery of consciousness follows a predictable course with CMD (diagnosed by an EEG motor imagery paradigm) being a transitory state from unconsciousness to emergence of consciousness. Characterizing the trajectory of recovery will be the overall goal of this application. We propose a two-center, observational cohort study of patients presenting with primary intracerebral hemorrhage (ICH) in the frontal lobe, thalamus, or striatocapsular region who do not have major bleeding in other regions (e.g., midbrain). We will divide ICH patients into two cohorts- conscious and unconscious. In the unconscious cohort, we will determine the time from injury to clinical command following with time to CMD as a time-varying covariate (Aim 1), and determine if CMD predicts long-term functional outcome (Aim 2). In the conscious cohort we will determine if patients with sensory aphasia are able to show command following when tested with the EEG motor imagery paradigm (Aim 3). To answer questions posed in Aims 1 and 2, we will study ICH patients that are unresponsive to commands. We will determine if location and volume of ICH and electrophysiological features on resting EEG are associated with time to CMD. We will determine if CMD independently predicts 6-month modified Rankin Scale after controlling for known predictors. To address Aim 3 we will study conscious ICH patients with and without sensory aphasia using the EEG motor imagery paradigm. This project will determine if CMD is a state that patients predictably transition through as they recover from an unconscious to a conscious state and quantify the false negative rate for command following in aphasic patients. These insights will fill an important gap in the understanding of impairment and recovery of consciousness of this common form of stroke that will likely be generalizable to other brain injuries.