Uncovering the Patient Perspective: An Analysis of 38,609 Social Media Posts Reveals Critical Gaps in Neurosurgical Communication

Introduction: Understanding patient experiences is crucial for improving neurosurgical care. While social media platforms offer unique insights into patient perspectives, comprehensive analyses of these discussions have been limited.

Methods: We analyzed 38,609 neurosurgery-related social media posts across 10 neurosurgery-related subreddits using natural language processing. We extracted 249,231 keyword matches across 185 unique clinical terms (mean=6.46 matches/post), analyzing sentiment patterns, term relationships, and discussion clusters along with network analysis.

Results: Analysis of 38,609 patient discussions revealed distinct patterns. Overall patient sentiment remained near-neutral (mean=0.020±0.412). Term association showed patients frequently discussed multiple related symptoms and concerns together (relationship density, 0.635; consistency, 0.830). "MRI" was among the most commonly referenced diagnostic (connectivity score=0.957) and served to connect different discussion themes (bridge score=0.012). Discussions about "normal" findings were the most central topic across all diagnostic conversations (centrality score=0.962). "Numbness" frequently appeared as the complaint associated with diagnostic workup (bridge score=0.012). Clustering identified 4 distinct discussion patterns: high anxiety discussions (sentiment=-0.283; n=9,652), neutral clinical discussions (-0.008; n=11,582), positive experience sharing (0.229; n=8,493), and symptom-focused discussions (-0.015; n=8,882). Notably, GBM-related discussions showed unexpectedly positive sentiment patterns (mean=0.102± 0.081; n=3,241). Further analysis revealed significant symptom associations: headache-migraine (P <.001), headache-intracranial pressure (P <.001), and confusion-migraine (P=.003). Common symptom groupings showed headache-pressure-vision changes (n=6,231; r=0.82), seizure-memory-cognitive changes (n=4,892; r=0.76), balance-coordination-gait difficulties (n=3,445; r=0.71), and pain-numbness-tingling (n=2,987; r=0.69). Patient sentiment was most negative related to diagnostic testing (sentiment=-0.412) and most positive discussing treatments (0.384).

Conclusion : This comprehensive analysis reveals that patient discussions form complex networks centered around diagnostic processes and symptom patterns rather than specific treatments. The high centrality of diagnostic-related terms suggests that diagnostic communication may be more important to patient experience than previously recognized. The identification of distinct discussion clusters with unique sentiment patterns could help structure patient education and support resources. Furthermore, the strong co-occurrence patterns between symptoms suggest that patients experience and discuss their conditions in more interconnected ways than typical clinical categorizations reflect. These findings have direct implications for structuring patient communication and support systems in neurosurgical practice.