Due to a change in data provider, the existing GRR platform is currently suspended. Future issues will focus on specific current topics in the fields of risk management and corporate finance.

For the latest information and insights into key risks, please contact timotej.jagric@um.si or aljaz.herman@um.si.

AI-powered decision support for risk-based VAT audit prioritisation

The VAT Audit Decision Intelligence Cockpit is an interactive research application developed at the Institute of Finance and Artificial Intelligence (IFAI), Faculty of Economics and Business, University of Maribor. The application translates advanced machine learning models into a practical decision-support environment for VAT audit prioritisation, demonstrating how artificial intelligence can improve public sector decision-making through economically informed resource allocation.

Unlike traditional approaches that focus solely on predicting the probability of detecting tax irregularities, the Decision Intelligence framework integrates multiple analytical components into a single decision process:

• estimation of the probability of detecting VAT irregularities,
• prediction of the expected additional VAT assessment,
• economic optimisation of limited audit capacity,
• transparent policy evaluation and scenario analysis.

Rather than maximising predictive accuracy alone, the system is designed to maximise the expected value of audit decisions, supporting more efficient allocation of inspection resources while maintaining full transparency of the underlying methodology.

The application enables users to:

• explore the complete Decision Intelligence workflow,
• compare different audit prioritisation strategies,
• analyse expected policy outcomes under different audit capacities,
• understand the economic rationale behind model recommendations,
• experiment with interactive "what-if" scenarios,
• evaluate decision quality using advanced metrics such as Financial Materiality Capture and Materiality Lift.

To ensure complete confidentiality, all demonstration data have been fully anonymised, with direct and indirect taxpayer identifiers removed or transformed. The application is intended exclusively for research, education and demonstration purposes and does not expose confidential taxpayer information.

This web application illustrates IFAI's broader research mission of transforming advanced artificial intelligence, quantitative modelling and Decision Intelligence into transparent and operational decision-support systems for finance, public administration and strategic management. It complements the Institute's ecosystem of browser-based analytical tools, including the Corporate Risk Management Composite Index (CRMI), the Industry Beta Monitor, the Beta Forecaster and the Global Risk Radar.

Access the application:

The Health System Decision Intelligence Platform is an advanced research application developed by the Institute for Finance and Artificial Intelligence (IFAI), University of Maribor. It demonstrates how artificial intelligence, explainable AI, digital twins, and scenario modelling can support evidence-informed decision-making for health systems and well-being economy policies.

Rather than providing another dashboard of indicators, the platform addresses a fundamental challenge facing governments and international organizations: how to transform evidence into better policy decisions. It moves beyond monitoring and benchmarking by enabling users to explore future development pathways, evaluate policy interventions before implementation, and assess the resilience of health systems under alternative scenarios.

The platform integrates several complementary Decision Intelligence modules:

• Country Trajectory Explorer – analyses the historical evolution of national health systems and their movement across the wellbeing landscape.
• Exact Scenario Laboratory – quantifies the expected impact of changes in individual health system characteristics.
• Alternative Pathways to Better Wellbeing – identifies multiple evidence-based reform strategies leading towards similar policy objectives.
• Health System Digital Twin – evaluates the resilience of health systems under alternative economic, demographic, epidemiological and policy scenarios.
• Explainable Artificial Intelligence – provides transparent explanations of model recommendations and identifies the structural drivers behind expected improvements.

The analytical core combines a Self-Organizing Map with an optimized Wellbeing Surface, allowing complex multidimensional relationships between health system characteristics and wellbeing outcomes to be represented in an intuitive visual environment. The methodology preserves the advantages of unsupervised machine learning while supporting scenario analysis, policy simulations, reform pathway exploration, and adaptive decision-making.

The platform has been designed as a prototype of a Health System Decision Intelligence Platform, illustrating how future policy support systems can evolve from measurement towards prediction, scenario evaluation, policy design, stress testing, and continuous learning. It is intended for researchers, policymakers, international organizations, ministries of health, public health agencies, and decision-makers interested in strengthening health systems and advancing the well-being economy.

Access the application: