ADVANCING DIGITAL TRUST IN ELECTION TRIBUNAL SYSTEMS THROUGH BLOCKCHAIN AND ARTIFICIAL INTELLIGENCE: TOWARD FORENSIC ACCOUNTABILITY AND SDG 16 IMPLEMENTATION

Abstract

This study evaluates the effectiveness of a Blockchain–AI Forensic Accountability Framewok (BAFAF) in election tribunal integrity using 2023 Lagos State gubernatorial election data on 20 Local Government Areas (LGAs) . Vote distribution analysis shows that the APC won 13 LGAs with a mean vote count of 39,800; the LP won 7 LGAs with a mean vote of 30,450, while the PDP recorded a consistent mean of 5,950 votes in each LGA. The turnout witnessed weathered from 14.8% to 24.6%. The three LGAs were flagged as anomalies with Z-score > 2.  The AI-based verification process produced normalized scores within the range of 0.91 to 1.00. As a result, seven LGAs were flagged for review. The 100% blockchain hashing of election result forms confirmed that all records were not tampered with. Machine learning evaluation results indicate that Isolation Forest achieved the best performance with an accuracy of 0.950, precision of 1.000, recall of 0.800, and F1-score of 0.889. In contrast, supervised models (Random Forest, Logistic Regression, SVM) recorded zero precision and recall because they were unable to detect anomalies due to the imbalanced class distribution within the dataset used. Regression analyses revealed significant coefficients for AI Verification Score (β = 0.671, p < 0.001) and Blockchain Verification Rate (β = 0.538, p < 0.001), while the total valid votes non-significant (β = 0.083, p = 0.112) as the evidence integrity strongest evaluators. Analysis of correlations supported the findings, with evidence showing a significant positive relationship with AI (r=0.891, p<0.01) and blockchain (r=0.844, p<0.01) metrics. The results reveal that the use of AI anomaly detection with blockchain verification can enhance transparency, traceability and forensic accountability of election tribunal evidence. Based on the real-world elections, this study shows that BAFAF is feasible, it was numerically shown that Isolation Forest is effective in detection and it provides an empirically founded mechanism to support the objective of SDG 16 on accountable, transparent and reliable institutions.