Decision making and planning have long relied on AI-driven forecasts, and the government and the general public are focused on minimizing risks and maximizing benefits in the face of future public health uncertainties. A recent study aimed to enhance forecasting techniques by utilizing the Random Descending Velocity Inertia Weight (RDV IW) technique, which improves the convergence of Particle Swarm Optimization (PSO) and the accuracy of Artificial Neural Network (ANN).
The RDV IW technique takes inspiration from the motions of a golf ball and modifies the velocities of particles as they approach the solution point. By implementing a parabolically descending structure, the technique aims to optimize the convergence of the models. Simulation results demonstrated that the proposed forecasting model, with a combination of alpha and alpha_dump values set at [0.4, 0.9], exhibited significant improvements in both position error and computational time when compared to the old model.
The new model achieved a 6.36% reduction in position error, indicating better accuracy in forecasting. Additionally, it showcased an 11.75% improvement in computational time, suggesting enhanced efficiency. The model reached its optimum level with minimal steps, showcasing a 12.50% improvement compared to the old model. This improvement is attributed to better velocity averages when speed stabilization occurs at the 24th iteration.
An important aspect of forecasting models is their accuracy performance. The computed p-values for various metrics, such as NRMSE, MAE, MAPE, WAPE, and R2, were found to be lower than the set level of significance (0.05). This indicates that the proposed algorithm demonstrated significant accuracy performance. Hence, the modified ANN-PSO using the RDV IW technique exhibited substantial enhancements in the new HIV/AIDS forecasting model when compared to the two previous models.
These findings suggest that the incorporation of the RDV IW technique can greatly improve the accuracy and efficiency of AI-driven forecasts. The optimization of convergence in models allows for better decision making and planning, especially in the context of public health uncertainties like HIV/AIDS. This study opens up possibilities for further research and applications of the RDV IW technique in other forecasting domains.