logologo

Popular Boards

Primary Link Pubmed DOI Openalex Full text (OA)

Key Takeaways

Sample Definition And Size

The study is a network meta‑analysis of randomized controlled trials (RCTs) evaluating six AI rehabilitation modalities (robot training [RT], brain‑computer interface [BCI], remote rehabilitation [RR], intelligent rehabilitation [IR], virtual reality [VR], and robot training combined with virtual reality [RT + VR]) in patients with upper limb dysfunction after stroke. A total of 101 RCTs involving 4,702 subjects were included (2,390 in experimental groups and 2,312 in control groups) ([doi.org](https://doi.org/10.3389/fneur.2023.1125172)).

Study Type

Network meta‑analysis of randomized controlled trials (RCTs), using both direct and indirect comparisons and SUCRA ranking to assess relative effectiveness of different AI rehabilitation interventions ([doi.org](https://doi.org/10.3389/fneur.2023.1125172)).

Conflicts Of Interest

No conflicts of interest were declared in the article (Conflict of interest section indicates none reported) ([doi.org](https://doi.org/10.3389/fneur.2023.1125172)).

Results Summary

Key findings: - FMA‑UE‑Total: No statistically significant differences between interventions; IR ranked highest (SMD = 0.02, 95% CI = –0.40 to 0.43; SUCRA = 70.5%) ([doi.org](https://doi.org/10.3389/fneur.2023.1125172)). - Modified Barthel Index (MBI): No significant differences; BCI ranked highest (SMD = 0.03, 95% CI = –0.24 to 0.29; SUCRA = 73.6%) ([doi.org](https://doi.org/10.3389/fneur.2023.1125172)). - FMA‑UE‑Proximal: RT + VR vs CT showed significant improvement (SMD = 0.43, 95% CI = 0.01 to 0.85); RT vs CT also significant (SMD = 0.32, 95% CI = 0.07 to 0.59); RT + VR ranked highest (SUCRA = 84.8%) ([doi.org](https://doi.org/10.3389/fneur.2023.1125172)). - FMA‑UE‑Distal: No significant differences; RT + VR ranked highest (SMD = 0.03, 95% CI = –0.47 to 0.52; SUCRA = 74.1%) ([doi.org](https://doi.org/10.3389/fneur.2023.1125172)). - Action Research Arm Test (ARAT): All AI modalities significantly improved ARAT vs CT: RT + VR (SMD = 0.73, 95% CI = 0.20 to 1.26), VR (SMD = 0.73, 95% CI = 0.14 to 1.32), BCI (SMD = 0.78, 95% CI = 0.25 to 1.31), RT (SMD = 0.93, 95% CI = 0.17 to 1.70), IR (SMD = 0.92, 95% CI = 0.36 to 1.48), RR (SMD = 0.91, 95% CI = 0.44 to 1.39); RT + VR ranked highest (SUCRA = 99.6%) ([doi.org](https://doi.org/10.3389/fneur.2023.1125172)). - Adverse effects: None reported across included studies ([doi.org](https://doi.org/10.3389/fneur.2023.1125172)).

Abstract

www.crd.york.ac.uk/prospero/#recordDetail, identifier: CRD42022337776.

Referenced In

Mercedes C.
2 months ago
AI Papers and Discussion

🧠 AI in Rehabilitation: The Hype vs. The Reality

New umbrella review of 32 systematic reviews by Abdalla et al. (2026) —here's what actually works and what's just noise.

🔑 The One Clear Win

Post-stroke upper limb recovery is thus far, the only area with reproducible evidence. A major network meta-analysis (Zhu et al., 2023) spanning 101 publications found robotic training + VR improves activity-level outcomes.

But here's the catch, any gains on impairment and daily independence vanish when assessors are blinded and practice dose is matched (Antoni et al., 2025)

  • Low back pain: AI-assisted physiotherapy also show no significant advantage over usual care (Kapil et al., 2025)

This simply means that the AI 'advantage' in its current state would be better described as 'comparable but not better' than conventional therapy.

⚠️The Brutal Lab-to-Clinic Drop

In the real world, conditions are messy: different hospitals, different equipment, different patient populations - Causing the actual implementation of AI to see some setbacks.

  • Brain-computer interfaces: ~99% offline accuracy → ~50% online in actual patients (Gutierrez-Martinez et al., 2021).

  • Computer vision for movement tracking? Falls apart under real-world conditions (Sardari et al., 2023).

💡 The Bottom Line for AI in Rehabilitation

Only stroke imaging AI is deployment-ready today. Everything else? Can be used as capacity extenders, but not pure replacements.

The writers close off with a demand for newer AI papers to provide Proof of meaningful functional gains, external validation, and equity-by-design before any adoption.

Read Full Post
2