Accelerating Referral for Thrombectomy in Acute Stroke Patients Using an Artificial Intelligence-based Software

Accelerating Referral for Thrombectomy in Acute Stroke Patients Using an Artificial Intelligence-based Software

The goal of this clinical trial is to learn if an AI-based imaging software, Methinks Stroke Suite, can reduce the time to transfer stroke patients who need thrombectomy from local stroke centers to specialized centers. The study focuses on acute stroke patients who are initially evaluated at local stroke centers that cannot perform endovascular therapy (EVT). The main questions it aims to answer are:

• Does Methinks Stroke Suite reduce the time it takes to decide if a patient needs to be transferred for thrombectomy?
• How accurate is the AI software in identifying patients who are candidates for EVT?

Researchers will compare the AI-based workflow to a historical cohort to see if the Methinks Stroke Suite improves transfer decisions and treatment times.

Participants will:

• Undergo a CT scan at the local stroke center, which will be analyzed by Methinks Stroke Suite.
• Be transferred to a thrombectomy-capable center if the AI + clinical judgment identifies them as potential EVT candidates.
• Be followed for 90 days after their stroke to assess recovery outcomes.

Study Overview:

The CATalyze-AI study aims to evaluate the impact of integrating the Methinks Stroke Suite, an artificial intelligence (AI) imaging software, into the workflow of stroke centers unable to perform endovascular therapy (EVT). This is a prospective, multicenter, quasi-experimental clinical trial focusing on acute stroke patients initially assessed at local stroke centers (Local-SC) in Catalonia, Spain. The introduction of this AI algorithm is expected to improve the decision-making process for transferring patients who may benefit from EVT, reducing time-to-transfer decisions.

Primary Objective:

The primary goal is to assess how much the Methinks Stroke Suite reduces the time from patient arrival at the Local-SC to the transfer decision (time-to-transfer decision) to a thrombectomy-capable center.

Secondary Objectives:

Accuracy of Methinks: Evaluate the software's accuracy in identifying large vessel occlusion (LVO) candidates requiring EVT.

Workflow Efficiency: Measure global workflow improvements, including door-in to door-out (DIDO) times and time to groin puncture.

Thrombectomy Rates: Assess population-wide EVT rates. Functional Outcomes: Evaluate patient recovery based on functional outcomes at 90 days post-stroke.

Safety Analysis: Monitor complications during patient transfers, especially in cases where Methinks does not suspect LVO.

Clinician Experience: Conduct a parallel study to assess the satisfaction of clinicians using the software.

Study Design:

This is a quasi-experimental, open-label trial. The Methinks Stroke Suite will be integrated into the usual stroke care workflow of participating centers, and the results will be compared with a historical cohort. Subjects will be followed for 90 days post-stroke to evaluate outcomes.

Study Setting and Population:

The study will take place across several centers in Catalonia, Spain, including:

Hospital Universitari Vall d'Hebron Hospital General de Granollers Consorci Hospitalari de Vic Hospital Clínic de Barcelona Centre Hospitalari Manresa - Fundació Althaia Mútua de Terrassa The target population is acute stroke patients evaluated at Local-SC. A sample size of 250 patients is projected. All patients meeting the inclusion criteria will be enrolled, and consent for data collection will be obtained at the destination center.

Eligibility Criteria:

Inclusion Criteria:

Suspected acute stroke. Patients aged 18 years or older, with no upper age limit. Stroke onset <24 hours since last known well. A non-contrast CT (NCCT) performed at the Local-SC and processed by the Methinks Stroke Suite.

Informed consent from the patient or legally designated representative.

Exclusion Criteria:

Patients in a coma (NIHSS > 1). Patients requiring life support due to unstable clinical status. Imaging that does not meet DICOM Tag and Acquisition Requirements.

Historical Cohort:

For time-reduction comparisons, a control group will be selected retrospectively from each participating center. These control patients will match study participants based on age, pre-stroke functional status, baseline NIHSS score, working hours (office vs. off-hours), and use of intravenous thrombolysis.

Intervention and Workflow:

When patients arrive at a Local-SC, they will be evaluated for suspected stroke and undergo a CT scan. Methinks Stroke Suite will analyze the imaging in real-time, and its results will be shared with the referring stroke neurologist. Based on the AI results and clinical judgment, patients suspected of LVO will be transferred to a thrombectomy-capable center.

In cases where Methinks does not detect LVO, CTA (CT Angiography) may be performed to reassess the patient's status. Transfer decisions will be based on clinical judgment and the results of this additional imaging.

Primary Outcome:

The primary outcome is the reduction in time-to-transfer decision, defined as the time from patient arrival at the Local-SC to the notification to Emergency Medical Services (EMS) for patient transfer to a thrombectomy-capable center.

Secondary Outcomes:

Accuracy of Methinks Stroke Suite in detecting LVOs. Improvement in workflow times, including the time from Local-SC arrival to groin puncture.

Increased EVT rates among patients flagged by Methinks as LVO positive. Safety outcomes, including complications during transfer. Functional outcomes based on the modified Rankin Scale (mRS) at 90 days. Clinician satisfaction with the use of Methinks Stroke Suite.

Sample Size and Power Calculation:

The study aims to include 152 patients to detect a 10-minute reduction in transfer decision time, with 80% power and a significance level of 0.05. To account for potential enrollment failures, the target is set at 250 patients.

Blinding and Statistical Analysis:

The trial is not blinded, as the intervention is visible to clinicians. However, neuroimaging and clinical data will be assessed by blinded reviewers to ensure unbiased evaluation. Statistical analysis will compare time-to-transfer decision between groups and measure accuracy (sensitivity, specificity) of Methinks predictions. Quantitative variables will be described using means and standard deviations, while qualitative variables will be analyzed using chi-square or Fisher's exact tests.

Data Management:

Data will be collected through existing stroke registries (CICAT, SITREM, TICAT, and SONIIA). Methinks Stroke Suite will automatically log patient imaging data. Blinded investigators will review the data to ensure it complies with the protocol.

Ethics and Regulatory Compliance:

The study follows European regulations on data protection and ethical standards. Informed consent will be obtained from all participants, and the Fundació Vall d'Hebron Institut de Recerca (VHIR) acts as the study sponsor.

Timeline:

The patient inclusion period is expected to last 18 months, followed by 3 months of follow-up, and 2 months for data analysis.

Inclusion Criteria:

1. Suspected acute stroke patients
2. >18 years with no upper age limit
3. Included < 24 h since last seen well
4. Non-contrast CT is available at the local stroke center
5. Image has been processed by Methinks Stroke Suite
6. Informed consent obtained from a patient or his or her legally designated representative (if locally required).

Exclusion Criteria:

1. Patients in a coma (NIHSS item of consciousness >1)
2. Patients with unstable clinical status who require emergent life support care
3. Subject imaging does not meet Image Acquisition and DICOM Tag Requirements