Orthopedic Research Highlights: Arthroscopy Utility and AI Prediction in Spine Surgery

A retrospective cohort study of pediatric patients undergoing medial patellofemoral ligament (MPFL) reconstruction without pre‑operative MRI findings found that 43% of 81 knees had intra‑articular pathology on diagnostic arthroscopy, and 12.3% required a change in surgical management, such as chondroplasty or loose‑body removal.

A systematic review and diagnostic test accuracy meta‑analysis of machine‑learning models predicting proximal junctional pathology after adult spinal deformity surgery pooled data from 16 studies (3,625 patients). Overall, models showed moderate discrimination with a pooled sensitivity of 0.51, specificity of 0.84 and an AUC of 0.67; the highest‑performing models improved sensitivity to 0.63 and AUC to 0.77. The authors conclude that current AI tools are useful for risk stratification but need better validation before guiding clinical decisions.