Data‐Driven Insights Into Psittacosis Transmission: Hybrid Mathematical‐Neural Network Framework With Real Epidemiological Data

Source abstract

ABSTRACT Psittacosis is a zoonotic disease primarily transmitted from infected birds or contaminated environments to humans, posing significant challenges for public health surveillance and poultry management. In this study, we propose a hybrid modeling framework that integrates a compartmental model with an artificial neural network (ANN) to analyze psittacosis transmission dynamics in both human and bird populations with environmental contamination. We first establish the fundamental dynamical properties of the model, positivity and boundedness of solutions, the basic reproduction number via next‐generation matrix method, and the stability of the disease‐free and endemic equilibria. Some parameters of the model are estimated using reported psittacosis data from Japan between 2002 and 2010 through Ordinary Least Squares fitting. The calibrated model is then used to generate simulated datasets, which serve as training data for the ANN. This allows the neural network to learn the underlying nonlinear dynamics governed by the psittacosis model while preserving epidemiological interpretability. Numerical simulations further examine the effects of bird transmission, environmental exposure, treatment, and preventive measures on disease persistence and control. The neural network predictions show strong agreement with the ODE‐generated solutions across a range of parameter values, confirming the reliability of the proposed framework for reproducing complex psittacosis dynamics.