Artificial intelligence in rehabilitation: a review of clinical effectiveness, real-world performance, safety, and equity across modalities and settings - Summary - MDSpire
Advertisement
Artificial intelligence in rehabilitation: a review of clinical effectiveness, real-world performance, safety, and equity across modalities and settings
To provide a comprehensive overview of artificial-intelligence–enabled rehabilitation and assess its clinical usefulness, safety, equity, and cost across various modalities.
Key Findings:
Technology-assisted training (robotics with or without VR) shows reproducible activity improvement for post-stroke upper limb rehabilitation.
Inconsistent effects on impairment and independence were noted when dose was matched and assessors blinded.
AI-enabled interventions often experience degradation in performance from development to deployment, particularly in brain-computer interfaces and computer vision.
Imaging-based decision support tools require local calibration and evaluation before clinical implementation.
Reported adverse events are generally mild, but usability, adherence, equity, and cost are under-researched.
Interpretation:
AI has the potential to enhance rehabilitation services but requires rigorous validation, safety measures, and equitable access to ensure effectiveness across diverse populations and settings.
Limitations:
Generalizability of AI models is a concern due to performance degradation across different environments.
Inconsistent safety and usability reporting, especially for home and hybrid rehabilitation pathways.
Lack of subgroup performance reporting and skewed representation towards high-income settings, leaving low-income populations underrepresented.
Conclusion:
AI can expand rehabilitation access and independence when adhering to clinical standards, with clear adoption criteria and ongoing monitoring to ensure safety, equity, and effectiveness.
by Nafisa Abdalla, Rabie Adel El Arab, Amany Abdrbo, Mohammed Almari, Mohammed Yahya Ayoub, Bilal Alsaaideh, Mohammad Suhail Dagamseh, Wesam Taher Almagharbeh, Fuad Abuadas, Mohammad S. Abu Mahfouz, Mastoura Khames Gaballah
