Related Experiment Video
Updated: May 13, 2026
07:35
Cognitive Function and Upper Limb Rehabilitation Training Post-Stroke Using a Digital Occupational Training System
Published on: December 29, 2023
Artificial Intelligence in Physical, Occupational and Neuro-Rehabilitation: Clinical Effectiveness, Prognostic
Rabie Adel El Arab1, Omayama Abdulaziz Al Moosa2, Wesam Taher Almagharbeh3
1Almoosa College of Health Sciences, Alhsa, 36422, Saudi Arabia. r.adel@almoosacollege.edu.sa.
Journal of Medical Systems
|May 12, 2026
Summary
Artificial intelligence (AI) in rehabilitation primarily supports existing programs rather than replacing therapists. Future scale-up requires robust data, validated models, and equitable access for trustworthy services.
Area of Science:
- Rehabilitation Medicine
- Health Informatics
- Artificial Intelligence in Healthcare
Background:
- Global rehabilitation demand outstrips capacity.
- Artificial intelligence (AI) offers potential for increased access, personalized therapy, and improved adherence.
- Evidence on AI's clinical effectiveness, prognostic capabilities, and implementation feasibility is needed.
Purpose of the Study:
- To synthesize evidence on the clinical effectiveness of AI-enabled rehabilitation.
- To evaluate the prognostic performance of AI models in rehabilitation.
- To assess the implementation feasibility of AI in diverse healthcare settings.
Main Methods:
- A mixed-methods systematic review was conducted.
- Searches included major biomedical and engineering databases (MEDLINE, Embase, IEEE Xplore, etc.).
- Methodological quality was assessed using MMAT and PROBAST+AI; evidence was synthesized thematically.
Main Results:
- Thirty diverse studies were included, showing modest and heterogeneous clinical effects.
- AI interventions were often comparable to conventional therapy, with limited long-term benefits.
- Prognostic models lacked external validation; implementation faced challenges in training, governance, and digital equity.
Conclusions:
- AI currently amplifies behavioral interventions in rehabilitation, not replacing therapist-delivered care.
- Scaling requires arm-symmetric adherence capture, validated outcomes, and transparent model deployment.
- Embedding equity, economic evaluation, and ongoing oversight is crucial for trustworthy AI services.
Keywords:
Artificial intelligenceDigital healthHealth equityImplementation scienceMHealthMachine learningPrediction modelsPrognostic modelsRehabilitationTelerehabilitation
