A Cooperative Automated Decision-Making Strategy with Fuzzy Contextual Modulation for SME Insolvency Prediction

Early insolvency prediction in Small and Medium-sized Enterprises (SMEs) is a critical challenge for financial stability, credit risk management, and public policy-making. Traditional approaches often rely on isolated predictive models that, despite achieving competitive performance, typically lack semantic integration, contextual adaptability, and operational transparency. This paper proposes a cooperative strategy for an Automated Decision-Making (ADM) system that integrates multiple survival analysis methods to estimate insolvency risk over a defined temporal horizon. The system is structured into three functional layers: (i) an input manager responsible for standardizing financial and non-financial indicators; (ii) a multi-model decision core based on heterogeneous survival architectures, whose outputs are harmonized through a semantic integration layer; and (iii) a fuzzy-based Contextual Modulator that calibrates technical risk estimates by modeling regulatory, sectoral, and socioeconomic criteria as fuzzy linguistic variables. This fuzzy logic approach allows the system to capture the inherent uncertainty and structural vulnerability of SMEs, incorporating contextual information beyond strict model-based risk estimates. The architecture is implemented through an interactive interface and incorporates Explainable Artificial Intelligence (XAI) techniques to ensure traceability, enhance interpretability, and facilitate the understanding of model outputs by decision-makers. The results show that the proposed cooperative strategy improves robustness through model cooperation compared with monolithic models and provides dynamic, fuzzy-calibrated risk curves that support contextualized intervention prioritization. This work contributes to the transition from isolated prediction models toward collaborative decision ecosystems aligned with the operational requirements of financial institutions and public policy organizations.