Real-time prospective (shadow mode) validation of an AI-based clinical decision support system for predicting 3-month functional outcome in acute stroke: the VALIDATE study protocol

Introduction Reperfusion therapies improve outcomes in acute ischaemic stroke, but hyperacute treatment decisions remain complex and are rarely informed by individualised outcome predictions. Artificial intelligence (AI)-based clinical decision support systems (CDSS) could provide real-time prognostic estimates, yet prospective evidence of their feasibility and performance within routine workflows is scarce. We aim to prospectively evaluate the real-time feasibility, usability and predictive performance of an AI…

By the Truvace deskOctober 17, 20261 min read1 sourcePeer-reviewedTRV-2026-1755· cite & verify

In brief

The VALIDATE protocol describes a prospective, multicentre observational study at three tertiary stroke centres. Consecutive patients with acute ischaemic stroke will be enrolled while treating physicians continue standard practice.

A researcher will enter patient data into the VALIDATE-CDSS app in real time, but the system will not influence clinical decisions. The study matters because evidence on AI prediction tools in routine stroke workflows is scarce. As a protocol, it reports no performance findings, so whether the tool is usable or accurate is still unknown.

Main points

The system predicts 3-month functional outcome on the modified Rankin Scale for four strategies: intravenous thrombolysis, endovascular thrombectomy, combined therapy or no reperfusion.
Predictions are generated at three time points: baseline clinical data, non-contrast CT and CT angiography.
The primary outcome is real-world feasibility and usability within the hyperacute stroke workflow.
Secondary outcomes include predictive performance, agreement between suggested and actual treatments, incremental value as data accumulate, and bias across predefined subgroups.

The gain

An AI decision support tool could give clinicians individualised 3-month outcome estimates for several stroke treatment options during hyperacute care, where such predictions are rarely available.

The rundown

The study design is observational and run in shadow mode. A dedicated researcher collects data and enters it through a mobile application, so clinical decisions are not influenced by the AI output.

Ethics approvals came from committees at Vall d'Hebron Institut de Recerca, Heidelberg Medical Faculty and Hadassah Medical Center-Ein Kerem, and the trial is registered as NCT05622539.

Results will be published in open-access journals and conference presentations. Anonymised data and metadata will be deposited in the Zenodo repository once the study is complete.