AI in Rehabilitation: An Umbrella Review of Clinical Effectiveness and Safety

An umbrella review synthesizing evidence on AI and device-based rehabilitation from biomedical, allied health, and engineering databases up to September 1, 2025. The 43.0 KB document, authored by Nafisa Abdalla and last updated in March 2026, distinguishes AI-enabled interventions from technology-assisted modalities. It analyzes outcomes across impairment, activity, independence, usability, safety, equity, and economics.

Use Cases

• Benchmarking clinical effectiveness of robotics and VR for post-stroke upper limb activity based on the review's synthesis.
• Assessing the lab-to-clinic performance drop for AI-enabled systems like brain-computer interfaces mentioned in the findings.
• Evaluating equity and subgroup fairness in rehabilitation AI based on the reported representation skew and under-measured metrics.
• Informing trial design and reporting standards based on the review's critique of contemporary AI standards in prediction models.

Strengths

• License is CC-BY-4.0, permitting broad reuse and adaptation.
• The review employs a structured Population–Exposure–Outcome framework for synthesis.
• Searches span multiple database types (biomedical, allied health, engineering) for comprehensive coverage.

Limitations

• The dataset is a 43.0 KB DOCX file, indicating a limited scope as a review document, not a primary data repository.
• Row count and column-level documentation are absent; the document's structure must be inferred after download.
• The description notes that usability, adherence, equity, and cost are under-measured in the underlying evidence.

Provenance

Source

figshare, authored by Nafisa Abdalla.

Collection Method

Umbrella review of reviews using a Population–Exposure–Outcome framework.

Time Range

Reviews from database inception to September 1, 2025.

Freshness

Last updated 2026-03-18 11:12:59