Abstract: Accurate prediction of deoxidation equilibria in Fe-based melts is essential for oxygen control and inclusion engineering, but remains difficult because conventional thermodynamic models are limited in describing both finite-concentration effects and strong short-range ordering in liquid alloys. In this work, a structure-informed coupled method (SICM) is developed by combining the Unified Interaction Parameter Formalism (UIPF) with the Modified Quasi-Chemical Model (MQM). In the proposed framework, weak interactions in the solvent-rich matrix are described by the UIPF, whereas strong solute-oxygen associations are represented by the MQM. Composition-dependent coordination numbers and binary mixing enthalpies are further introduced to define the atomic pairing sequence and coupling weights. The model is validated for the Fe-Al-O and Fe-Si-O systems using 193 experimental datasets at 1873 K and is further extended to the Fe-Cr-Al-O system. Compared with the evaluated thermodynamic models, the SICM gives the lowest prediction errors for the Fe-Al-O and Fe-Si-O systems and maintains comparable accuracy for the Fe-Cr-Al-O system, with mean absolute percentage errors of 10.0%, 6.2%, and 20.5%, respectively. The results indicate that incorporating structural information into a coupled thermodynamic framework improves the description of deoxidation behavior over a wide composition range.