Assessing the Impact of Quarantine Behavior Among Individuals Infected with Mpox: A Fractional Artificial Neural Network Approach

This study proposed a combination of compartmental modeling with a bio-inspired machine learning (ML) framework for the dynamical analysis of monkeypox (MPOX) transmission. A Caputo fractional-order model is developed based on human infection data from the USA to understand the long-term fluctuation in disease propagation and inherent memory effects. The transmission dynamics of MPOX incorporating vaccination, quarantine, recovery, loss of immunity, and both human to human and animal to human transmission pathways are considered. The fractional-order model is numerically solved using the Adams-Bashforth-Moulton (ABM) method to investigate the effects of fractional order on disease dynamics. Global sensitivity analysis based on Latin hypercube sampling and partial rank correlation coefficient (LHS-PRCC) analysis identifies the key parameters influencing the human reproduction number (R2). To provide an efficient surrogate for the numerical solutions, an artificial neural network (ANN) based on the Levenberg-Marquardt backpropagation (LMB) algorithm is developed. The dataset obtained through ABM is used to reduce the mean square error (MSE) with the statistics of 70% as training and 15% for both testing and validation. The results indicate that human to human transmission is a major driver of the epidemic, whereas timely quarantine and vaccination substantially reduce transmission. The fractional-order formulation further shows that decreasing the fractional order slows epidemic progression and delays the infection peak, highlighting the influence of memory effects on disease dynamics. The proposed fractional-order and neural network framework provides an efficient and reliable approach for eradicating the infectious diseases transmission within the population and for assessing the effectiveness of key intervention strategies.