Scientific Reports: Article in Press
Transforming Credit Risk Evaluation in Digital Lending from Black Box Models to Transparent Decisions
Anber Abraheem Shlash Mohammad, Suleiman Ibrahim Mohammad, Asokan Vasudevan, S. M. Ferdous Azam, Lakshmi Sevukamoorthy, Manoranjan Parhi, M. Ugarthi Shankalia & Zaeid Ajsan Salami
Received: 2 December 2025 | Accepted: 29 April 2026 | Published online: 13 May 2026
Executive Impact: Key Metrics & Breakthroughs
Our analysis reveals how integrating optimization and interpretable AI models leads to unparalleled performance and clarity in digital lending, directly impacting your bottom line.
0
Optimized Classification Accuracy
0
Explainable AI Confidence
0
Cross-Validation Stability
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Digital Lending & BNPL Landscape
The rapid expansion of Buy Now, Pay Later (BNPL) services has democratized credit access but challenged traditional risk assessment. Many BNPL users lack extensive financial histories, making conventional scoring models inadequate. Furthermore, existing machine learning solutions often act as "black-box" models, hindering interpretability and compliance.
Proposed AI Framework
This study introduces an optimization-driven hybrid machine learning framework designed for robust and transparent credit risk prediction. It integrates gradient boosting models (LightGBM, CatBoost, EBM) with nature-inspired metaheuristic optimization (Brown Bear Optimization Algorithm, Puma Optimizer). The approach leverages systematic data preprocessing, handles class imbalance, and employs feature engineering to extract meaningful patterns from loan applications, ensuring transparency through feature-level explanations.
Performance & Interpretability
The model was rigorously evaluated using multiple performance metrics across cross-validation folds, demonstrating improved stability and predictive capability. Specifically, optimized configurations showed significantly higher accuracy and F1-scores compared to baselines. Feature-level explanations, derived from SHAP analysis, identified key variables like 'agel_credit' as most influential, balancing predictive power with critical interpretability for regulated financial environments.
Real-World Impact
The framework's methodological synergy, enhanced interpretability, and robust performance make it suitable for deployment in regulated financial sectors. It not only minimizes portfolio risks and fosters financial inclusion but also offers a transparent, accountable decision-making process. Its cross-domain applicability extends to healthcare, insurance, and engineering, demonstrating potential for broad strategic impact by transforming complex data into understandable, actionable insights.
99.50%
Max Test Accuracy Achieved by Optimized CAPO Model
Enterprise Process Flow
Data Pre-Processing
→
Model Development
→
Model Evaluation
→
Model Calibration and Interpretability
→
Insights and Decision Support
→
Deployment & Practical Use
| Feature | Baseline LGBM | Optimized LGPO |
|---|---|---|
| Accuracy (All Data) | 0.9002 | 0.9802 |
| F1 Score (All Data) | 0.9018 | 0.9803 |
| Key Advantages of Optimized Models |
|
|
Beyond Finance: Broad AI Applicability
Challenge: Many industries, like healthcare, insurance, and engineering, require precise, clear, and stable predictions, but often face limitations with traditional black-box AI models that lack transparency and adaptability for diverse, high-stakes scenarios.
Solution: The proposed framework offers methodically optimized and interpretable models, delivering exact, clear, and stable predictions. This allows organizations to move beyond financial applications, making robust data-driven decisions in critical operational contexts.
Impact: Businesses can reduce operational and financial risks, make better decisions in complex environments, and transform raw data into understandable, actionable insights. This leads to the creation of reliable, high-value prediction systems across multiple domains, enhancing both strategic and ethical outcomes.
Quantify Your AI Advantage
Use our interactive calculator to estimate the potential time and cost savings for your enterprise by implementing an advanced, interpretable AI framework.
Estimated Annual Cost Savings
$0
Hours Reclaimed Annually
0
Your AI Implementation Roadmap
A structured approach ensures seamless integration and maximum impact for your enterprise.
Phase 1: Discovery & Strategy
Collaborative workshops to define project scope, data availability, and desired business outcomes. Establish key performance indicators (KPIs) and success metrics tailored to your organizational goals. This phase typically lasts 2-4 weeks.
Phase 2: Data Engineering & Model Prototyping
Collect, clean, and transform your enterprise data, building robust data pipelines. Develop and prototype initial AI models, including the integration of interpretable boosting and metaheuristic optimization techniques. Focus on feature engineering and validation. This phase spans 6-10 weeks.
Phase 3: Model Refinement & Interpretability Tuning
Iteratively optimize model hyperparameters, focusing on balancing accuracy, stability, and interpretability. Implement SHAP and permutation importance analyses to ensure decision transparency and regulatory compliance. Rigorous cross-validation is performed here. This phase lasts 4-8 weeks.
Phase 4: Integration & Deployment
Integrate the validated AI models into your existing enterprise systems and workflows. Develop monitoring dashboards for real-time performance tracking and operational alerts. Conduct pilot programs and user training. This phase typically takes 8-12 weeks.
Phase 5: Continuous Optimization & Scaling
Establish a feedback loop for continuous model improvement, adapting to evolving data and business needs. Explore opportunities for scaling the AI solution across other departments or use cases, maximizing long-term ROI and competitive advantage. Ongoing support and maintenance.
Ready to Transform Your Enterprise with Transparent AI?
Leverage cutting-edge interpretable AI to drive smarter, more reliable decisions across your organization. Book a free, no-obligation consultation with our AI strategists.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat