Accurate soil moisture monitoring is critical for precision agriculture. This study proposes a hybrid approach for water stress detection, integrating classic hydro-physical parameters (Field Capacity and Permanent Wilting Point) with machine learning algorithms. Using volumetric moisture (????) data from capacitive sensors collected in a Red-Yellow Latosol, a Decision Tree classifier was trained to validate and automate the identification of stress episodes. The results demonstrated that the computational model converged to the same thresholds defined by laboratory analyses, achieving an accuracy greater than 99%. This highlights the viability of cyberphysical systems that combine the precision of hardware sensors with the decision robustness of artificial intelligence for irrigation management.