AI-supported triage in emergency departments

Emergency departments at CHwapi, Saint-Luc and ZOL together handle over 200,000 visits a year. All three use the Manchester Triage System at the desk, and all three saw the same pattern in their 2025 audits: under-triage of older patients with atypical presentations, and over-triage during peak hours when the triage nurse has under three minutes per patient.

The call asked for decision support that runs alongside the triage nurse, uses structured data and free-text notes in Dutch and French, and gives an explainable second opinion rather than an autonomous score. We asked for TRL 5 or higher, a description of the training data, and a plan for silent-mode validation before any live use.

Eleven companies submitted between February and April. Seven were invited to pitch on the Friday afternoons in April and May; four were forwarded internally as commercial products outside the pre-commercial focus. After the feedback meeting of 25 June, two solutions were selected for a retrospective validation on anonymised 2025 triage data from the three hospitals, supported by an Innoviris application led by Saint-Luc and an EIT Health application led by ZOL.

The call is closed. Results of the retrospective validation are expected in the first quarter of 2027 and will be presented at a members day. Companies working on emergency-flow solutions can still submit through the regular intake.