Smart-SABI: Digital Phenotyping of Stroke Access Barriers (Smart-SABI)

Machine Learning Identification of Modifiable Access Barriers in Acute Ischemic Stroke: A Multimodal "Digital Phenotyping" Approach

This study aims to identify and quantify the non-clinical barriers (social, transport, and knowledge-based) that delay patient arrival at the hospital during an Acute Ischemic Stroke. By utilizing a multimodal approach that combines a validated patient questionnaire (SABI Tool), Geographic Information Systems (GIS) analysis, and biological markers (infarct volume), the investigators seek to develop a Machine Learning model capable of predicting high-risk phenotypes for pre-hospital delay. The ultimate goal is to validate "Social Determinants of Health" against objective biological outcomes.

Study Overview

• Status: Enrolling by invitation
• Conditions: Stroke; Healthcare Disparities; Thrombectomy; Practice; Healthcare Access
• Intervention / Treatment: Behavioral: Targeted Stroke Systems of Care Training (SABI-Guided)

Detailed Description

Despite advances in stroke reperfusion therapies (thrombectomy and thrombolysis), pre-hospital delays remain the primary cause of preventable disability. Current triage systems rely heavily on clinical severity scales but fail to account for Social Determinants of Health (SDOH) that dictate onset-to-door times.

This is a prospective, observational, single-center cohort study designed to validate the "Stroke Access Barrier Identification" (SABI) tool using a "Triangulation Strategy."

The study employs three distinct data sources:

Subjective: Administration of the SABI questionnaire to assess cognitive, physical, and structural barriers.

Geospatial (Objective): Network-based GIS analysis to calculate precise drive-time isochrones and public transit density, validating patient reports of transport difficulty.

Biological (The "Anchor"): Correlation of barrier scores with Infarct Core Volume (measured via CT-Perfusion/MRI) and 90-day functional outcomes.

Data will be processed using interpretable Machine Learning algorithms (Random Forest / XGBoost) and SHAP (SHapley Additive exPlanations) values to identify the specific social features that most strongly predict delayed presentation and increased brain tissue loss.

• Study Type: Observational
• Enrollment (Estimated): 250

Contacts and Locations

Study Locations

• Egypt
  • Alexandria, Egypt
    • Alexandria Stroke and Neurointervention Center

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: Yes

• Sampling Method: Non-Probability Sample

Study Population

The source population comprises patients recruited from 12 tertiary centers across the MENA region: Alexandria University, Ain Shams University, and Cairo University (Egypt); Eskisehir Osmangazi University and Dr. Lutfi Kirdar City Hospital (Turkey); Amman Specialized IR Center (Jordan); King Khalid University, King Abdullah Medical City, and Imam Abdulrahman Al Faisal University (Saudi Arabia); Institute National de Neurology (Tunisia); Cleveland Clinic Abu Dhabi (UAE); and Weill Cornell Medicine (Qatar).

This multinational design ensures significant geographic heterogeneity-ranging from the dense urban traffic of Istanbul and Cairo to the mountainous terrain of Abha-which is critical for GIS transport analysis. Additionally, the inclusion of diverse economic and cultural backgrounds supports robust SABI analysis regarding stroke awareness and health-seeking behaviors across the region.

Description

Inclusion Criteria:

• Diagnosis of Acute Ischemic Stroke (AIS) confirmed by neuroimaging (CT or MRI). Age $\geq$ 18 years. Presentation to the Emergency Department within 7 days of symptom onset (to ensure recall accuracy).

Patient or Legally Authorized Representative (LAR) able to provide informed consent.

Verifiable residential address (required for GIS analysis).

Exclusion Criteria:

• In-hospital stroke onset. Stroke mimics (e.g., seizure, complex migraine, hypoglycemia). Hemorrhagic stroke. Homelessness or lack of fixed address (precludes geospatial analysis). Severe aphasia or cognitive deficit without an available surrogate/caregiver to complete the questionnaire.

Study Plan

How is the study designed?

Number of groups / cohorts

1

Cohorts and Interventions

Group / Cohort

Intervention / Treatment

Acute Ischemic Stroke (AIS) Patients; Acute Ischemic Stroke (AIS) Patients This cohort consists of adult patients presenting to the Emergency Department with a confirmed clinical and radiological diagnosis of Acute Ischemic Stroke. The group encompasses a continuous spectrum of arrival times, subsequently stratified during analysis into "Early Arrivers" (presenting within the therapeutic window, typically < 4.5 hours) and "Late Arrivers" (presenting after the therapeutic window).

Behavioral: Targeted Stroke Systems of Care Training (SABI-Guided); Implementation of targeted barrier-reduction strategies at selected stroke centers based on baseline SABI profiles. The primary intervention consists of EMS Training Programs focused on stroke recognition, triage protocols, and rapid transport to Mechanical Thrombectomy (MT) capable centers. Comparator/Control: Pre-intervention period (historical control) where standard of care was utilized without the targeted SABI-guided training. Post-Intervention: Assessment of MT utilization rates and SABI scores following the implementation of the training modules.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Correlation of SABI Score with Infarct Core Volume (The Biological Anchor)	To validate if subjective barriers correlate with objective physiological damage. The total score on the SABI questionnaire (Scale 0-100, higher scores indicate higher barriers) will be correlated with the admission Infarct Core Volume (measured in milliliters via automated CT-Perfusion software).	Baseline (Admission Imaging)

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Predictive Accuracy of ML Model for "High-Risk" Delay	Sensitivity and Specificity of the XGBoost Machine Learning model in classifying patients as "Early Arrivers" vs. "Late Arrivers" (defined as >4.5 hours from Last Known Well) using combined clinical and SABI variables.	Baseline through Study Completion (12 months)
Agreement between Subjective Transport Barriers and GIS Metrics	Cohen's Kappa coefficient measuring agreement between patient-reported "Difficulty with Transport" (SABI Domain 2) and objective "Network Drive Time" calculated via ArcGIS using historical traffic data.	Baseline
Functional Outcome (mRS) at 90 Days	Correlation between baseline SABI Barrier Score and the Modified Rankin Scale (mRS) score at 90 days. The mRS is a scale from 0 (no symptoms) to 6 (dead).	90 Days post-discharge

Collaborators and Investigators

• Sponsor: Middle East North Africa Stroke and Interventional Neurotherapies Organization

Study record dates

Study Major Dates

• Study Start (Actual): October 11, 2025
• Primary Completion (Estimated): February 11, 2027
• Study Completion (Estimated): April 11, 2027

Study Registration Dates

• First Submitted: November 20, 2025
• First Submitted That Met QC Criteria: November 20, 2025
• First Posted (Actual): December 2, 2025

Study Record Updates

• Last Update Posted (Actual): December 2, 2025
• Last Update Submitted That Met QC Criteria: November 20, 2025
• Last Verified: November 1, 2025

More Information

Terms related to this study

• Keywords: SABI; thrombectomy access; Stroke Access Barriers Index
• Additional Relevant MeSH Terms: Cerebrovascular Disorders; Brain Diseases; Central Nervous System Diseases; Nervous System Diseases; Vascular Diseases; Cardiovascular Diseases; Stroke
• Other Study ID Numbers: MENASINO105; NALAregistrySABI2026 (Registry Identifier: NALAregistrySABI ; Digital Phenotyping of Stroke Access Barriers)

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: UNDECIDED

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No