| Objective To explore the use of artificial intelligence time series clustering methods to identify pain progression trajectories in patients with knee osteoarthritis (KOA) and analyze baseline demographic, psychological, and biomechanical differences across these trajectories. Methods The study was based on the publicly available Osteoarthritis Initiative (OAI) database, including participants with complete or imputed knee injury and osteoarthritis outcome score (KOOS) pain scores for their right knee at baseline, 12, 24, 36, and 48 months. Pain scores were converted into pain intensity indicators (Pain intensity=100-Koos pain Score). Dynamic time warping combined with clustering algorithms was used for unsupervised clustering. The final number of clusters was determined by a composite scoring approach. Differences between groups were assessed using analysis of variance (ANOVA) or rank-sum tests, with post-hoc comparisons adjusted for multiple testing. Results A total of 4,345 participants were included. The composite score suggested two clusters, identifying two distinct pain progression trajectories (1,771 participants in one group and 2,574 in the other). Baseline differences showed significant variations in sitting-stand time (F=33.152, P＜0.001), quadriceps strength (F=8.249, P=0.004), 400-meter walking time (F=6.860, P=0.009), smoking pack-years (F=6.019, P=0.014), and depression scores (F=5.871, P=0.015) between the trajectories. The two trajectories also showed significant differences in the distribution of imaging grades for knee osteoarthritis (χ2=13.000, P=0.011). Post-hoc comparisons indicated that the group with more rapid progression had higher depression scores, more pronounced structural changes, greater smoking burden, and lower muscle strength. Conclusion Artificial intelligence time series clustering effectively identifies two major pain progression trajectories in KOA, revealing dynamic differences in pain evolution. Patients with sustained worsening of pain are more likely to experience depression, muscle weakness, and functional limitations, while those with better function and muscle strength tend to show greater pain improvement. |
