This paper presents an innovative method based on Artificial Neural Networks (ANN) for optimizing the design of microstrip antennas using HFSS software. The method leverages a dataset generated from comprehensive full-wave electromagnetic simulations to accurately predict antenna performance. In the ANN model, coefficients of reflection as input parameter, and five output parameters representing key geometrical variables of the antenna are utilized. The study demonstrates how ANNs effectively predict these variables, thereby enhancing design efficiency and significantly reducing the computational time required for antenna optimization. By providing a robust and efficient framework, the results highlight the promising potential of ANNs in optimizing antenna designs for diverse applications, including medical and wireless communication systems. Specifically, the proposed Ultra-Wideband (UWB) antenna operates over a frequency range from 5.17 GHz to 7.02 GHz and maintains an S11 value of -20.49 dB at 5.8 GHz within the Industrial, Scientific, and Medical (ISM) bands. This research contributes valuable insights into advancing electromagnetic performance through efficient antenna design methodologies.