Modern Credit Risk Modelling: From Classical Scorecards to Deep Learning

Credit risk modeling has long been a cornerstone of financial stability, underpinning lending decisions across banking, insurance, and capital markets. Built on a foundation of statistical rigour, regulatory compliance, and interpretability, classical approaches such as logistic regression, survival analysis, and scorecards have shaped both academic research and real-world risk management practice for decades. The landscape is now changing fundamentally. The proliferation of alternative data sources, including high-frequency transaction records, behavioral signals, financial narratives, and unstructured textual information, is redefining what it means to assess creditworthiness. At the same time, advances in machine learning and deep learning are delivering predictive capabilities that challenge established modeling paradigms, enabling more granular, dynamic, and forward-looking risk assessments across retail, corporate, and sovereign portfolios.Despite these advances, significant challenges remain. Reconciling model complexity with transparency, satisfying evolving regulatory frameworks such as Basel IV and IFRS 9, addressing fairness and bias in automated decisions, and maintaining robustness under economic stress and concept drift all demand multidisciplinary solutions. This Research Topic provides a platform to advance AI-enabled credit risk modeling across theory, methodology, and practice.The goal of this Research Topic is to address the pressing need for intelligent, interpretable, and regulatory-aligned credit risk models in an era of data abundance and algorithmic complexity. Traditional approaches are no longer sufficient to capture the richness of modern financial data, the interconnectedness of credit networks, or the pace of macroeconomic change. By integrating advances in machine learning, deep learning, and generative AI with established credit risk principles, this collection aims to develop rigorous methodologies for default prediction, loss estimation, stress testing, and responsible automated decision-making. The Research Topic seeks to foster collaboration between academia, industry, and regulators, enabling scalable, explainable, and compliant AI-driven credit risk solutions.This Research Topic welcomes contributions related to:• Interpretable Machine Learning, Constrained Models & Hybrid Scorecard Frameworks• Deep Learning, Sequential Modeling & Temporal Credit Risk Analytics• Graph Neural Networks & Interconnected Financial Risk Modeling• Alternative Data, NLP & Multimodal Credit Assessment• Foundation Models & Large Language Models for Credit Risk• Fairness, Bias Mitigation & Algorithmic Accountability in Lending• Explainability, Counterfactual Reasoning & Actionable Decision Support• Model Risk Management, Regulatory Compliance & Basel/IFRS Alignment• Stress Testing, Scenario Simulation & Generative Models• Dynamic Risk, Concept Drift & Macroeconomic Adaptation Topic editors: Flavio Barboza, Herbert Kimura, Marcelo Botelho Da Costa Moraes Source: https://www.frontiersin.org/research-topics/81565/modern-credit-risk-modelling-from-classical-scorecards-to-deep-learning