An artificial flora algorithm-based recurrent neural network for multi-disease prediction in healthcare monitoring system

Background: The growing patient population in today's digitally transformed healthcare environment is pushing the industry toward more advanced remote medical monitoring solutions. The capacity of these systems to accurately foresee health conditions is essential to their e ectiveness. While widely used, conventional illness prediction models—which primarily depend on machine learning techniques—often have poor predictive accuracy. Method: Our study presents a novel solution to this shortcoming: an improved Recurrent Neural Network (RNN) model that is further optimized by using the Arti cial Flora (AF) technique. Through the careful use of the AF algorithm to optimize the weight parameter of the RNN, this novel technique greatly improves the model's performance in multi-disease prediction. Findings: Our study includes a thorough assessment and contrast of the suggested RNN-AF model with other well-known models, such as the RNN and RNN- Particle Swarm Optimization (RNN-PSO). Key performance parameters including accuracy, sensitivity, and specificity provide a solid foundation for this comparison research. Interpretation: Our tests' empirical results highlight the RNN-AF model's better performance and demonstrate how well it outperforms other models in terms of illness prediction accuracy.