Predictive factors for walking in acute stroke patients: a multicenter study using classification and regression tree...

Area of Science:

• Neurology
• Rehabilitation Medicine
• Clinical Prediction Modeling

Background:

• Predicting walking ability in acute stroke is essential for timely rehabilitation.
• Current bedside assessments lack a comprehensive predictive model for independent ambulation.

Purpose of the Study:

• To develop and validate a prediction model for independent walking in acute stroke patients.
• Utilize classification and regression tree analysis for model construction.

Main Methods:

• Multicenter case-control study with 240 acute stroke patients.
• Data collected: NIH Stroke Scale, Brunnstrom Recovery Stage (lower extremities), ability to turn over from supine, and higher brain dysfunction.
• Functional Ambulation Categories used to define independent (FAC ≥ 4) and dependent (FAC ≤ 3) walkers.

Main Results:

• A prediction model was developed using Brunnstrom Recovery Stage, ability to turn over, and higher brain dysfunction.
• Four patient categories identified with distinct independent walking probabilities (0% to 82.5%).
• Model effectively stratified patients based on predicted walking outcomes.

Conclusions:

• A validated bedside prediction model for independent walking in acute stroke patients was established.
• The model, based on three key criteria, offers a practical tool for early clinical decision-making.
• Facilitates personalized and efficient rehabilitation strategies for stroke survivors.