Evaluating Patient Satisfaction Through Validated Predictive Classification Models

Patient satisfaction plays a crucial role in evaluating healthcare service quality. This study explores the field of predicting patient satisfaction in healthcare settings. The research aims to develop a predictive and validated model for patient satisfaction scores to improve patient satisfaction through simulation-optimization. Factors related to patient satisfaction were identified, including length of stay (LOS) and patient sociodemographic characteristics. A survey was conducted to collect relevant data. Several predictive modeling techniques were employed, i.e., ordinal logistic regression, decision trees, gradient boosted trees, and neural networks. The models were trained on the survey data to predict patient satisfaction scores. A simulation was then used to generate synthetic data, which was fed into the predictive models to obtain predicted satisfaction scores. The predicted scores were validated against the actual survey scores using the Mann-Whitney U test to evaluate each model's ability to represent the simulated system accurately. Findings suggest that ordinal logistic regression provides the best performance for predicting satisfaction across various LOS. The validated predictive model can be integrated into a simulation-optimization framework to identify strategies for improving patient satisfaction in healthcare systems, facilitating informed decision-makina and the allocation of resources.