Table 3.

Clinical outcomes comparison of different injection methods.

Method	First-pass puncture success rate	VAS score reduction (6 weeks)	ASES score improvement (3 months)	Re-injection rate (12 months)	References
Blind injection	65 ~ 70%	1.0 ~ 1.5 points	15 ~ 20%	35 ~ 40%	Aly et al. (2015) and Enriquez et al. (2025)
Ultrasound-guided injection	80 ~ 85%*	1.8 ~ 2.3 points*	25 ~ 30%	20 ~ 25%	Aly et al. (2015) and Huang et al. (2015)
AI-Navigated ultrasound injection	90 ~ 95% (inferred#)	2.2 ~ 2.8 points (inferred#)	30 ~ 35% (inferred#)	10 ~ 15% (inferred#)	Huang et al. (2015), Andriollo et al. (2024), and Kuratani et al. (2022)

*Direct evidence: Cited RCT/meta-analysis data for blind injection and conventional ultrasound-guided injection (e.g., 80 ~ 85% first-pass success rate, 1.8 ~ 2.3-point VAS reduction). These data are supported by consistent clinical observations across diverse patient cohorts, including paraplegic patients with rotator cuff tendinopathy.

#Inferred data: Clearly noted that AI-navigated injection outcomes (90 ~ 95% first-pass success rate, 2.2 ~ 2.8-point VAS reduction) are derived from technical principle extrapolation and analogous procedure data. The inference is based on two core technical rationales: (1) AI’s precise anatomical segmentation (Dice coefficient > 0.9 for rotator cuff muscles) reduces off-target injections; (2) real-time needle tracking and adaptive path planning address the limitations of static ultrasound guidance in dynamic injection processes. However, these inferred data lack validation from dedicated RCTs focusing on AI-navigated shoulder injections, and future clinical studies are required to confirm these outcomes.