Development and Validation of a Prognostic Model for Neurocritical Patients Using Multimodal Brain Monitoring

This study is a prospective, observational, multicenter cohort study conducted in China. It focuses on neurocritical patients with acute brain injury, including large cerebral infarction, intracranial hemorrhage, subarachnoid hemorrhage, or severe traumatic brain injury. The study will collect multimodal data from EEG, TCD, and NIRS monitoring within 1-3 days of ICU admission. Machine learning algorithms will be used to develop a model predicting 90-day outcomes, categorized as awake, comatose, or deceased. The study population includes patients aged 18-80 years. Exclusion criteria include pre-existing severe neurological disorders, incomplete monitoring data, or withdrawal during the study period.

Inclusion criteria

• Aged 18-80 years, no gender restrictions.
• Diagnosed with acute brain injury (ABI), including one of the following: large cerebral infarction, supratentorial large-volume intracerebral hemorrhage, subarachnoid hemorrhage, or severe traumatic brain injury, with imaging evidence (CT or MRI) supporting the diagnosis.
• On ICU admission, Glasgow Coma Scale (GCS) eye response = 1 (no eye opening) and motor score ≤ 5 (does not follow commands); or within 48 hours, neurological deterioration with no eye opening and motor score reduced to ≤ 5 (total GCS score ≤ 8).
• Able to undergo continuous multimodal monitoring, with an expected ICU stay of ≥72 hours.
• Informed consent signed by the family or legal representative.

Exclusion criteria

• Confirmed brain death on admission or imaging showing irreversible brain herniation.
• Severe trauma unrelated to brain injury (e.g., multiple fractures, spinal cord injuries, or visceral rupture) that may interfere with brain function monitoring or outcome assessment.
• Pre-existing severe neurological disorders such as epilepsy, severe encephalopathy, or chronic intracranial conditions (e.g., brain tumors or hydrocephalus).
• Inability to perform multimodal monitoring due to technical issues (e.g., equipment failure or sensor installation problems).
• Predicted survival time <24 hours after admission, or family members choose to withdraw treatment.
• Refusal to participate in the study by the patient or their legal representative.