Social Network Analysis and Behavioral Insights to Enhance Patient Engagement in Chiropractic Care: A Quasi-Experimental Study Using Virtual Reality and Deep Learning

Background: Chiropractic care is widely applied for pain relief and musculoskeletal improvement, yet limited emotional support and one-way communication may reduce trust and engagement. This quasi-experimental study aimed to examine the effect of a mobile Augmented Reality Application (ARA), combined with a Convolutional Neural Network (CNN) predictive model, on enhancing patient satisfaction, adherence, and clinical outcomes within chiropractic care.Methods: A pre-test post-test quasi-experimental design was implemented from January to December 2022 in a private chiropractic clinic in Mashhad, Iran. Out of 73 eligible patients with musculoskeletal disorders, 51 completed the study, with low back pain and neck pain being the most common issues reported. Participants were randomly assigned to an intervention group (n=21; standard care + ARA app) or control group (n=30; standard care only). Outcomes included functional performance (mental health, vitality, social functioning, disability, and kinesiophobia), treatment satisfaction, and adherence. Assessments were conducted at baseline and weeks 3, 5, and 8 using validated tools, including the 12-item Short Form Health Survey Questionnaire (SF-12) and the Fear Avoidance and Belief Questionnaire (FABQ). The satisfaction rate was assessed at week 8 using a 10-item questionnaire developed by the research team, while treatment adherence was monitored by app usage in the intervention group or clinic attendance in the control group.Results: By the eighth week, the intervention group reported higher satisfaction (91.6±6.5) compared to the control group (63.3±12.5; p<0.01). Significant improvements were observed in functional performance measures, including mental health, vitality, social functioning, and fear avoidance (p<0.01). A positive correlation was also found between adherence and satisfaction (r=0.402, p=0.003). The CNN model demonstrated moderate predictive accuracy with a Root Mean Squared Error (RMSE) of 0.1121 and a correlation coefficient of 0.1491 (p<0.01).Conclusion: The ARA app significantly improved outcomes, suggesting a scalable, patient-centered digital strategy. However, further extensive and long-term trials are recommended to validate its scalability.