Computer-vision detection of delirium from ICU patients' facial expressions

Research on Delirium Recognition in Neurocritical Patients Based on Facial Expression Behavior Patterns

This project tests whether a smartphone app using computer vision can detect delirium from facial expressions in adult neurocritical care patients.

Quick facts

• Study type: Observational
• Enrollment: 1000 (estimated)
• Ages: 18 Years to 80 Years
• Sex: All
• Sponsor: Beijing Tiantan Hospital (Academic / other)
• Locations: 1 site (Beijing, Beijing Municipality)
• Trial ID: NCT07136207 on ClinicalTrials.gov

What this trial studies

This prospective cohort uses a mobile application to record dynamic facial videos of neurocritical ICU patients following a CAM-ICU–based stimulus paradigm and DSM‑5 delirium diagnosis by an experienced specialist. Recordings are taken twice daily during ICU admission (8:00–10:00 AM and 8:00–10:00 PM) and captured within five minutes of the clinical delirium assessment. Images are preprocessed, augmented, and fed into a convolutional neural network (VGG16) pipeline to construct multidimensional behavioral features and build a classification model. The study was approved by the Beijing Tiantan Hospital Ethics Committee and combines clinical input with engineering expertise from the Institute of Computing Technology, CAS.

Who should consider this trial

Good fit: Adults (over 18) admitted to a neurocritical care ICU—including postoperative neurosurgical and stroke patients—who can provide informed consent or have proxy consent.

Not a fit: Patients with persistent coma, facial paralysis or major facial disfigurement, severe dementia, Parkinson's disease, or very short ICU stays are unlikely to benefit from facial-recognition–based detection.

Why it matters

Potential benefit: If successful, the system could help clinicians detect delirium earlier and more consistently by providing automated facial-expression analysis.

How similar studies have performed: Some pilot and small-scale studies applying computer vision and AI to delirium or emotion detection have shown promise, but large-scale validation in neurocritical populations is limited.

Eligibility criteria

Inclusion Criteria:

1. Neurocritical patients admitted to the ICU, including postoperative neurosurgical patients, stroke patients, and those receiving ICU care due to other neurological conditions.
2. Age over 18 years.
3. Signed informed consent.

Exclusion Criteria:

1. Age under 18 years.
2. Persistent coma (GCS ≤ 8) within 7 days pre- and post-surgery, making delirium assessment impossible.
3. Did not survive more than 24 hours in the ICU.
4. Patients with facial paralysis, post-traumatic facial disfigurement, or other conditions that could significantly affect facial recognition.
5. Exclusion of patients with severe dementia, Parkinson's disease, depression, or other conditions that might impact facial emotional expressions.

Where this trial is running

Beijing, Beijing Municipality

• Beijing Tiantan Hospital — Beijing, Beijing Municipality, China (Recruiting)

Study contacts

• Study coordinator: Huang Huawei, Doctoral degree

Conditions Delirium; Artificial Intelligence; neurocritical care units; Facial recognition