E-CAM-S — Electroencephalographic Confusion Assessment Method Severity Score

What this tool is and how to interpret its evidence

E-CAM-S is an automated research method that estimates delirium severity from quantitative frontal electroencephalography. It is not a patient questionnaire or a rapid bedside screen. The study calculated signals from four frontal bipolar EEG channels and extracted time- and frequency-domain measurements from repeated six-second epochs. Summary statistics across epochs produced 1,192 candidate inputs, which were reduced within a penalised learning-to-rank model. The resulting continuous score ranges from 0 to 1 and was trained to follow the ordering of clinical CAM-S values from 0 to 7. Higher values represent greater physiological abnormality associated with delirium severity; no practice-ready diagnostic threshold was established.

The retrospective study analysed 373 adults referred for EEG because of altered mental status at one tertiary academic hospital. Ten-fold patient-level internal cross-validation was used. E-CAM-S correlated with CAM-S at 0.68 in the full cohort, falling to 0.52 after comatose patients were excluded. Its area under the curve for separating non-delirious from delirious patients was 0.70. E-CAM-S was also associated with hospital mortality and length of stay.

This is promising physiological-model evidence, not external clinical validation. The selected cohort included many comatose patients, clinical CAM-S assessments came from multiple providers, and the clinical comparator was not an independent gold standard. Dementia, aphasia and non-English speakers were excluded, and sedative or other drug effects were not modelled directly. EEG availability and processing also limit routine use. E-CAM-S therefore belongs in the emerging physiological context register, not among ordinary bedside tools.