Non-Invasive Eye Tracking for the Diagnosis of Delirium on ICU

• Conditions: Delirium

• Registration Number: NCT04589169
• Lead Sponsor: Chelsea and Westminster NHS Foundation Trust

Brief Summary

In this trial, the investigators seek to demonstrate the feasibility of a system in continuously detecting 'inattention' as a subset of CAM-ICU in a small representative sample of patients in the Intensive Care Unit (ICU) at Chelsea \& Westminster NHS Foundation Trust.

Detailed Description

Delirium is an acute confusional state that affects many patients admitted to the hospital, especially intensive care. The current diagnosis of delirium is through the use of the Confusional Assessment Method in Intensive Care Unit (CAM-ICU) task based questionnaire. The core prinicipal to CAM-ICU is inattention; this is tested through asking the patient to remember a task and execute it on demand, e.g. squeezing the operator's hand everytime the letter A is said and then spelling CASABLANCA.

The aim of this study is to find correlates to inattention. Eye-gaze data is ideally suited for this task as eyes move to pay attention to the environment.

A video camera based eye-tracker has been developed that sits at the end of the bed (head-camera) and another behind the patient (scene-camera). The head-camera uses machine learning to measure the gaze of the patient's eyes while the scene-camera finds what the patient is looking at. Simulations are then run from the scene camera and the patient's gaze is then compared to find whether the patient is paying attention to what is simulated.

Once per day, a member of the local research team will fill in a non-validated questionnaire based on work by MacMurchy et al.

M. MacMurchy, S. Stemler, M. Zander, C. P. Bonafide, Acceptability, Feasibility, and Cost of Using Video to Evaluate Alarm Fatigue, Biomedical Instrumentation \& Technology 51 (2017) 25-33.

Study Timeline

• Study First Submitted: 2020-10-09
• Study First Submitted QC: 2020-10-09
• Study First Posted: 2020-10-19
• Status Verified: 2020-12
• Study Start: 2020-12-07
• Last Update Submitted: 2020-12-07
• Last Update Posted: 2020-12-08
• Primary Completion: 2021-10-12
• Study Completion: 2021-10-12

Recruitment & Eligibility

• Status: UNKNOWN
• Sex: All
• Target Recruitment: 30

Inclusion Criteria

1. Aged >18
2. Predicted Delirium as defined by the Early PREdiction of DELIRium in ICU patients (E-PREDELIRIC) score ≥ 20%
3. Expected length of stay ≥ 2 days

Exclusion Criteria

1. Non-concordant eyes
2. Visual Impairment
3. Dementia
4. Inability for facial recognition and eye tracking to be performed reliably
5. Lack of signed consent form / nominated consultee form
6. In-ability to perform CAM-ICU reliably

Study & Design

• Study Type: OBSERVATIONAL
• Study Design: Not specified

Primary Outcome Measures

Name	Time	Method
Demonstrate eye gaze correlates with CAM-ICU	8-12 months	Patient's usual clinical care will not be affected by the assessments performed by this study. In our institution, CAM-ICU is performed at least twice daily and more so if there are changes to mental status. The system will be switched on 5 minutes prior to performing CAM-ICU and will record for a duration of 10 minutes following which, the system will be turned off and the cameras covered. This procedure will occur every time CAM-ICU is performed.

Secondary Outcome Measures

Name	Time	Method
Acceptance of the use of cameras and artificial intelligence on ICU	8-12 months	Once per day, the research staff will complete an acceptability questionnaire designed to ascertain the acceptability of the use of cameras within this study on day to day activities in ICU.This questionnaire is based on work by MacMurchy et al.; * Minimum value is 0, maximum value is 4.; * Higher scores indicate that the system affected the nurse's ability to communicate with patients, nurses and doctors and that it affected the nurse's ability to care for them.; * A yes answer = 1 point, a No answer = 0 points. Marking will be done per question to gain insights into how the system affected the nurses.

Trial Locations

Locations (1)

Chelsea and Westminster Hospital NHS Foundation Trust; 🇬🇧 London, United Kingdom