Artificial Intelligence-Driven Comparison of Platelet-Rich Plasma and Hyaluronic Acid Efficacy in Knee Osteoarthritis Treatment: A Prospective Randomized Controlled Trial

Background: Despite the widespread use of platelet-rich plasma (PRP) and hyaluronic acid (HA) in knee osteoarthritis (KOA) management, their comparative efficacy remains controversial. This study aimed to systematically evaluate their therapeutic effects through artificial intelligence-based analysis of clinical and molecular outcomes. Methods: In this prospective, double-blind, randomized controlled trial, 140 patients with KOA were randomly assigned to receive either PRP (n = 70) or HA (n = 70) treatment. Clinical assessments included WOMAC, KOOS, and VAS scores, while molecular evaluation focused on inflammatory biomarkers (IL-1β, TNF-α, hs-CRP). A deep learning model was developed to predict treatment outcomes and identify key predictive factors. Primary outcomes were assessed at baseline, 2, 6, and 12 months post-treatment. Results: PRP demonstrated superior therapeutic efficacy across multiple parameters. At 12-month follow-up, the PRP group showed significantly greater improvements in pain reduction (SMD = 0.82; 95% CI: 0.45–1.19; P < 0.001) and inflammatory marker suppression (IL-1β reduction: 51.2%, P < 0.001) compared to the HA group. The artificial intelligence model achieved exceptional predictive performance (AUC = 1.000) in treatment outcome prediction. IL-1β emerged as the strongest predictor of treatment response (SHAP value: 0.185, P < 0.001), followed by WOMAC scores (0.175, P < 0.001). Subgroup analysis revealed optimal model performance in patients under 60 years (accuracy: 90.5%, 95% CI: 88.2%-92.8%). Discussion: Our findings extend beyond previous studies by demonstrating PRP's superior efficacy through both clinical and molecular evidence. The marked reduction in inflammatory markers, particularly IL-1β, suggests that PRP's therapeutic mechanism involves significant immunomodulation. The developed AI model's exceptional predictive accuracy offers a promising tool for treatment optimization, while the identification of key predictive biomarkers provides new insights into patient stratification. The enhanced efficacy in younger patients and those with moderate disease severity suggests optimal timing for PRP intervention, though the slightly lower prediction accuracy in severe cases indicates the need for careful patient selection. Conclusions: This study provides robust evidence for PRP's superior efficacy in KOA treatment through comprehensive clinical and molecular analyses. The developed artificial intelligence model demonstrates remarkable potential for personalizing treatment selection, with inflammatory markers serving as key predictive indicators. These findings advance the field of personalized medicine in osteoarticular diseases by establishing a framework for precision treatment planning.