Predicting patient satisfaction in digital orthodontics using a KAC based model integrating AI support and risk stratification

Understanding and predicting patient satisfaction in digital orthodontics is essential for ensuring treatment effectiveness and optimizing human–AI collaboration. However, structured behavioral models for satisfaction prediction remain limited. This study aimed to develop and validate a patient satisfaction prediction model based on the Knowledge–Attitude–Challenge (KAC) framework in digital orthodontics, and to evaluate its discriminative performance and clinical utility in risk stratification. A cross-sectional survey was conducted among 500 orthodontic patients and 500 orthodontists across 31 provinces in China. Two KAC-based questionnaires were developed and validated to quantify the three dimensions. High satisfaction was defined as a score of ≥ 4 on item Q6. A multivariate logistic regression model was constructed and evaluated using area under the ROC curve (AUC), calibration curve, and decision curve analysis (DCA). Patients were stratified into low-, medium-, and high-risk groups based on predicted probabilities. Higher Knowledge (OR = 1.15, 95% CI: 1.09–1.21) and Attitude scores (OR = 1.10, 95% CI: 1.06–1.15) were positively associated with high satisfaction, whereas Challenge scores were negatively associated (OR = 0.94, 95% CI: 0.90–0.98). The model demonstrated good discriminatory power (AUC = 0.79) and yielded notable clinical net benefit within the threshold range of 0.2–0.6. Satisfaction rates varied significantly across risk strata: 89.1% in the low-risk group versus 51.3% in the high-risk group (P for trend < 0.001). The KAC-based model effectively predicts patient satisfaction in digital orthodontics and offers strong interpretability and clinical relevance. It may assist clinicians in identifying patients at risk of low satisfaction and support the optimization of AI-assisted treatment pathways and patient-centered communication strategies.