Comprehensive Evaluation of Time Series Models for Urban Traffic Flow Prediction: A Comparative Study of ARIMA, GARCH, Prophet, and LSTM Approaches

Abstract: The proposed research presents the design and implementation of an intelligent fuzzy system that integrates disease prediction with personalized drug dosage control. The framework utilizes patient-specific data, including clinical, demographic, and physiological parameters, to predict disease probabilities and recommend safe drug doses through fuzzy inference mechanisms. By addressing uncertainties and vagueness in medical data, the system improves diagnostic reliability and ensures therapeutic accuracy. Comparative results show that the fuzzy-based approach achieves superior performance over conventional machine learning models such as ANN and SVM, obtaining 92% accuracy with a significantly lower RMSE of 0.19. The proposed system demonstrates strong potential for clinical decision support by enhancing interpretability, reliability, and precision in diagnosis and dosage recommendation.

Keywords: Fuzzy logic, disease prediction, drug dosage control, medical decision support, fuzzy inference system, machine learning, intelligent healthcare, uncertainty modeling.