Stroke
Conditions
Keywords
Stroke, Modified Rankin Scale, Artificial Intelligence
Brief summary
This study aims to validate the clinical performance of an artificial intelligence (AI)-based automatic scoring system for the Modified Rankin Scale (mRS). The core comparison is the consistency and accuracy between the AI-generated scores and standardized manual mRS follow-up assessments performed by trained professionals. The goal is to provide a convenient, efficient, and objective tool for stroke prognosis assessment, reduce the subjective variability of manual scoring, and optimize the stroke follow-up workflow.
Detailed description
This is a prospective, multicenter, observational study designed to validate the diagnostic performance of an AI-based automated scoring system for the Modified Rankin Scale (mRS) in patients with stroke. The primary objective is to evaluate the agreement between AI-generated mRS scores and standardized manual assessments conducted by trained clinicians. Secondary endpoints include the system's sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) in classifying functional outcomes.
Interventions
None listed
Sponsors
Xuanwu Hospital, Beijing
Capital Medical University
People's Hospital of Beijing Daxing District
Beijing Tiantan Hospital
The First Hospital of Fangshan District,Beijing
Study design
Observational model
COHORT
Time perspective
PROSPECTIVE
Eligibility
Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
No
Inclusion criteria
1. Age ≥ 18 years, male or female. 2. Clinically diagnosed with stroke, and confirmed by cranial CT/MRI to have stroke. 3. Clinically stable, with basic communication ability at discharge or outpatient visit. The patient or a fixed family caregiver is able to cooperate with telephone follow-up at 1 week after discharge or outpatient visit. 4. Signed informed consent by the patient or their legally authorized representative.
Exclusion criteria
1. Neurological deficits caused by non-stroke etiologies (e.g., brain tumor, traumatic brain injury, encephalitis). 2. Presence of severe disturbance of consciousness, severe cognitive impairment, psychiatric disorders, or global aphasia at discharge/outpatient visit, preventing effective communication; neither the patient nor family can cooperate with follow-up or assessment. 3. Combined with severe multi-organ failure (e.g., cardiac, hepatic, renal, respiratory), with an expected survival of less than 1 month, making completion of the 1-week follow-up impossible. 4. Long-term bedridden without a fixed caregiver, with no confirmed contact for follow-up, or refusal to participate in telephone follow-up and mRS assessment. 5. Incomplete clinical data, preventing baseline data collection.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Agreement Between Artificial Intelligence (AI)-Based and Manual Modified Rankin Scale (mRS) Assessments | 7 days post-discharge or post-outpatient visit, ± 2 days | The weighted kappa coefficient quantifies the level of agreement between the Artificial Intelligence (AI)-generated Modified Rankin Scale (mRS) scores and the standardized manual mRS assessments performed by trained clinicians |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Agreement Between AI-based and Manual Assessments of Dichotomized Modified Rankin Scale (mRS) | 7 days post-discharge or post-outpatient visit, ± 2 days | The simple kappa coefficient quantifies the level of agreement between the Artificial Intelligence (AI)-generated dichotomized Modified Rankin Scale (mRS) scores (0-2 vs. 3-6) and the standardized manual mRS assessments performed by trained clinicians |
| Bland-Altman Limits of Agreement Between AI and Manual Modified Rankin Scale (mRS) Scores | 7 days post-discharge or post-outpatient visit, ± 2 days | The Bland-Altman limits of agreement analysis evaluates the consistency between the Artificial Intelligence (AI)-generated and manually assessed Modified Rankin Scale (mRS) scores. The difference between manual and AI scores will be plotted on the y-axis against their mean on the x-axis, with limits of agreement (mean difference ± 1.96 × standard deviation) calculated. The analysis aims to visually assess how agreement varies across the range of mRS scores and identify any proportional bias, such as greater disagreement in patients with severe disability. |
| Diagnostic Performance of AI-Based vs. Manual Modified Rankin Scale (mRS) Dichotomization | 7 days post-discharge or post-outpatient visit, ± 2 days | The diagnostic performance analysis evaluates the ability of the Artificial Intelligence (AI)-based Modified Rankin Scale (mRS) scoring system to classify functional outcomes, using manual assessment as the reference standard. A 2×2 contingency table will be constructed for the dichotomized mRS categories (good outcome: 0-2 vs. poor outcome: 3-6). The analysis will calculate sensitivity, specificity, positive predictive value, negative predictive value, accuracy, and Youden's index. A receiver operating characteristic (ROC) curve will be plotted, and the area under the curve (AUC) will be computed. |
Contacts
CONTACTQingfeng Ma, MD
CONTACTZixin Wang, MD Candidate
STUDY_CHAIRQingfeng Ma, MD
Xuanwu Hospital, Beijing
Outcome results
None listed