A design of predictive computational network for transmission model of Lassa fever in Nigeria

This paper presents an innovative artificial neural networks (ANNs) based hybrid algorithm of genetics optimization and sequential quadratic programming (AGOSQP) to construct the mathematical model for the dynamics of Lassa fever (DLF) in Nigeria. The model designated by the transmission of disease between two populations: human population i.e. susceptible (Sh), exposed (Eh), infectious (Ih) and recovered (Rh) humans and rodent population i.e. susceptible (Sr) and infectious (Ir) rodents. The log sigmoid function as an objective function based on mean squared error is constructed to optimize AGOSQP where genetic algorithm work as global searching optimization and SQP serve as the local searching optimization. To assess the correctness, robustness and convergence stability, the comparison between state of art Adam method and proposed AGOSQP is established. The Theil's inequality coefficient (TIC), root mean square error (RMSE) and mean absolute deviation (MAD) are also computed to authenticate the efficiency of proposed AGOSQP to solve the model for the dynamics of Lassa fever.