An Improved Genetic Algorithm-Optimized Back Propagation Neural Network for Acrophobia Prediction Model

Acrophobia is one of the most prevalent phobias, significantly impacting patients’ daily lives. To address this issue, this study developed a predictive model for acrophobia using an improved Genetic Algorithm (IGA) optimized Back Propagation Neural Network (BPNN). By collecting relevant factors for diagnosing acrophobia, the BPNN was employed to uncover the inherent connections among these factors and establish the predictive model. In order to overcome the limitations of traditional neural networks, which often fall into local optima due to random assignment of initial weights and thresholds, this study refined the fitness selection method, crossover probability formula, and mutation probability formula of the GA. Furthermore, we optimized the initial weights and thresholds for the BPNN, thereby enhancing the predictive accuracy of the model. The study involved 930 subjects, with the complete dataset divided into training (70%) and test (30%) sets. Subsequently, a five-fold cross-validation was conducted within the training set to ensure the robustness of the model’s performance. To validate the predictive accuracy of the IGA-optimized BPNN model, this paper conducted several simulation experiments as the control group, with clinical diagnosis by psychiatrists serving as the reference standard. The results indicated that the IGA-optimized BPNN model can predict acrophobia with significantly better accuracy than the control group.