Operational intelligence for institutional processes: dynamic modeling and state-based decision policies

The management of research processes in higher education institutions is characterized by fragmented operational flows, manual validations, and static decision criteria, resulting in cumulative latency and limited adaptive capacity amid changing workloads. Although artificial intelligence technologies have been incorporated into institutional management, their application remains focused on isolated automation or predictive analysis, without explicitly integrating events, states, and decisions within a unified operational structure. This work proposes an operational intelligence framework in which institutional processes are modeled as a coupled dynamic system formalized as S _t+1 = T(S _t , E _t , D _t ). From operational logs, state trajectories are reconstructed, and two decision policies are evaluated: a deterministic rule-based baseline and a multivariable state-dependent policy. The evaluation is performed using operational metrics oriented toward system dynamics, including decision rate, activation variance, state sensitivity, and transition behavior. The results show that the proposed policy reduces the decision rate from 0.1427 to 0.1002 and the activation variance from 0.3689 to 0.1903, while increasing state sensitivity to 1.0. Under high-load conditions, the model generates transitions with average magnitudes of -777.79, compared to -60.17 in the baseline formulation, while short trajectories exhibit highly selective interventions with stabilization rates of 1.0. These results demonstrate a reconfiguration of the operational decision regime through selective state-dependent interventions.