Predicting Upper Limb Motor Impairment Recovery after Stroke: A Mixture Model

Area of Science:

• Neurology
• Rehabilitation Medicine
• Biostatistics

Background:

• Spontaneous recovery of upper extremity function after stroke is crucial.
• The 70% proportional recovery rule for the Fugl-Meyer motor upper extremity (FM-UE) scale is widely used but has limitations in predicting recovery.
• There is a need for more precise models to understand and predict individual recovery patterns.

Purpose of the Study:

• To develop a longitudinal mixture model for predicting Fugl-Meyer motor upper extremity (FM-UE) recovery after stroke.
• To identify distinct subgroups of patients based on their FM-UE recovery trajectories.
• To internally validate the predictive accuracy of the developed model.

Main Methods:

• Developed an exponential recovery function incorporating subgroup probabilities, proportional recovery coefficients (r k), time constants (τ k), and initial FM-UE score distributions.
• Fitted the model to FM-UE data from 412 first-ever ischemic stroke patients.
• Performed cross-validation to assess endpoint prediction accuracy and cluster assignment.

Main Results:

• Identified 5 distinct subgroups with varying recovery parameters (r k and τ k).
• The model predicted FM-UE endpoints with a median absolute error of 4.8 at 1 week and 4.2 at 2 weeks post-stroke.
• Achieved high accuracy (0.79-0.81) in assigning patients to poor, moderate, or good recovery clusters within the first two weeks post-stroke.

Conclusions:

• Upper extremity recovery after stroke is heterogeneous, with distinct patient subgroups exhibiting unique recovery profiles.
• The developed longitudinal mixture model provides accurate predictions of FM-UE recovery endpoints and cluster assignments.
• These findings enhance the understanding of post-stroke upper limb recovery patterns within the initial six months.