Stroke rehabilitation works best when it follows the rules of motor learning: high repetition, task-specific practice, and appropriate challenge. The problem is that most stroke patients receive nowhere near the dose required for meaningful cortical reorganization.

Why CIMT works

Constraint-induced movement therapy (CIMT) is the only approach with consistent evidence for inducing lasting cortical reorganization. The mechanism is not just forcing use of the affected limb—it is massed practice that drives competitive plasticity, expanding motor cortex representation of the trained movements at the expense of compensatory strategies.

The original CIMT protocol from Taub et al. requires 6 hours of daily training for 2 weeks, plus a mitt on the unaffected hand to prevent compensatory use. Subsequent trials (EXCITE, i-EXCITE) have modified this to make it more practical, but the core principle remains: thousands of repetitions are needed to drive structural changes in motor circuits.

What the dose problem looks like

Most stroke rehabilitation delivers 1-2 hours of therapy per day, 3-5 days per week. This is not enough. Wolf et al. showed that even 3 hours of CIMT-style training produces smaller effects than the full protocol. The motor system needs massed practice to engage competitive plasticity mechanisms.

The result is that patients plateau earlier than they should. Compensatory movements—using trunk rotation to reach instead of arm extension—become entrenched because they work well enough for daily tasks. Once these patterns are established, they are harder to unlearn.

Brain stimulation as adjunct

Non-invasive brain stimulation shows promise but not magic. Repetitive TMS to the contralesional hemisphere can reduce interhemispheric inhibition that suppresses the injured motor cortex. tDCS may prime motor cortex for learning during subsequent training.

The effect sizes are modest—roughly equivalent to 10-20% faster recovery when added to standard therapy. The key is that stimulation is not a replacement for training; it is a primer that makes training more effective. Trials that test stimulation alone generally fail.

What actually works

1. Subacute high-intensity training: The AVERT trial showed that very early mobilization (within 24 hours) can be harmful, but starting high-intensity training in the subacute phase (days to weeks post-stroke) improves outcomes.

2. Task-specific practice with high repetition: Robotics and gaming systems can help deliver the necessary dose. The key is not the technology—it is the repetition count.

3. Aerobic exercise: Cardiovascular fitness promotes BDNF release and general brain health. Post-stroke exercise programs improve not just fitness but also cognitive and motor outcomes.

4. Addressing depression: Post-stroke depression independently predicts poor motor recovery. SSRI treatment for depression may have the side benefit of improving recovery through serotonergic effects on plasticity, though the evidence is mixed.

The unsolved problem

About 30% of stroke patients have little to no motor recovery regardless of intervention. This group may have damage to critical motor pathways—particularly the corticospinal tract—that plasticity cannot overcome.

Identifying who falls into this group early remains a clinical challenge. Advanced imaging (DTI of the corticospinal tract) can help, but is not widely available. For these patients, diverting to compensatory training rather than restoration may be the better approach.

Testable predictions

1. Modified CIMT protocols delivering 4+ hours daily will outperform standard care, but 2-hour protocols will not show significant differences
2. rTMS or tDCS combined with high-dose training will outperform training alone, but stimulation alone will not
3. Early identification of corticospinal tract integrity will predict who benefits from restoration-focused versus compensation-focused rehabilitation

Limitations

Most evidence comes from mild-to-moderate stroke patients with some preserved hand function. Severe strokes are understudied. The dose-response relationship for different impairment levels is not well established.

Research synthesis from stroke rehabilitation literature including CIMT trials, EXCITE and i-EXCITE studies, AVERT trial, and rTMS meta-analyses.