Deep neural network driven prediction and control of dengue induced thrombocytopenia in paediatric patients using a fractional order mathematical model

Abstract

This paper presents an exhaustive fractional-order mathematical model for dengue-induced thrombocytopenia in children. The model contains asymptomatic carriers and seven epidemiological compartments that correspond to various levels of infection stages and platelet depletion. A Caputo fractional derivative is employed to account for the memory effect in disease development, which corresponds to long-term physiological dependencies and the effect of previous infection states on present dynamics. Important clinical parameters like the infection transmission rate, cure efficacy, and patient response to recovery are integrated to assess their effect on the outcome of disease. The existence and well-posedness of solutions are proven using fixed-point theory under the fractional-order model. The Newton polynomial-based iterative interpolation method is employed for numerical approximation to solve the system. Different treatment and therapeutic intervention methods are compared to evaluate effectiveness in dengue spread control and alleviation of platelet loss severity. The fractional-order model is also combined with a deep neural network to enhance predictive ability, and it performs well on training, testing, and validation datasets. The integration of fractional calculus and machine learning provides an all-encompassing solution for deciphering paediatric dengue dynamics and tailoring treatment regimens to improve recovery.