Abstract: Predicting renewal demand in old residential areas is critical for effective urban planning, particularly in addressing deteriorating infrastructure and maximizing resource distribution. Conventional statistical models often fail to capture complex and non-linear relationships among socio-economic, institutional, and environmental factors influencing residents' decision-making. This study utilizes an Artificial Neural Network (ANN) to predict renewal demand and determine the most influential factors (MIF), based on a dataset of 800 responses and twelve input factors classified into building involvement (BI), trust in government (TG), and local government image (LGI). The best ANN architecture demonstrated strong prediction performance, evidenced by a correlation coefficient of R=0.9523, Mean Squared Error (MSE) = 0.0192, and Mean Absolute Percentage Error (MAPE) = 3.4837%, signifying good accuracy and generalization capacity. Sensitivity analysis employing Garson's algorithm (GA) indicated that trust in government capability, fairness in execution, and decision transparency are the MIFs, followed by emotional attachment and place identity. The results emphasize that renewal demand is primarily driven by institutional trust and psychological attachment rather than physical conditions. These findings establish a data-driven basis for developing more responsive, inclusive, and sustainable urban renewal initiatives.