Explainable AI for Root Cause Analysis of Missing Data: Beyond Detection to Insight

Most current research—like Escudero-Arnanz et al.’s tensor completion for heart failure or Wang et al.’s z-score anomaly detection—focuses on identifying where data are missing or anomalous, but not why these gaps exist. Inspired by the call for transparency in Bernard Owusu Antwi et al. (2024) and the critique of poor missing data reporting in Yu et al. (2023), this research would develop an explainable AI framework that not only detects missing data but also traces its root causes. By incorporating model interpretability tools (e.g., SHAP, LIME) and linking anomalies with contextual metadata (device malfunctions, human workflows, environmental disruptions), auditors, health professionals, and engineers could gain actionable insight into systemic weaknesses or process failures. This approach addresses the “incidental findings” heuristic—turning missingness from a nuisance into an information-rich signal for system improvement. The impact? Organizations could proactively address root causes of data absences, improving data integrity and trust across domains from finance to healthcare.

References:

1. Enhancing audit accuracy: The role of AI in detecting financial anomalies and fraud. Bernard Owusu Antwi, Beatrice Oyinkansola Adelakun, Damilola Temitayo Fatogun, Omolara Patricia Olaiya (2024). Finance & Accounting Research Journal.
2. A Methodology to Detect Traffic Data Anomalies in Automated Traffic Signal Performance Measures. Bangyu Wang, Grant G. Schultz, Gregory S. Macfarlane, Dennis L. Eggett, Matthew C. Davis (2023). Future Transportation.
3. Low-Rank Tensor Completion for Heart Failure Exacerbation Detection in Multivariate Time Series with Missing Data. Óscar Escudero-Arnanz, Rosa Sicilia, Cristina Soguero-Ruíz, Inmaculada Mora-Jiménez, Diana Lelli, Claudio Pedone, Antonio G. Marques (2024). 2024 IEEE 37th International Symposium on Computer-Based Medical Systems (CBMS).
4. What is Missing in Missing Data Handling? An Evaluation of Missingness in and Potential Remedies for Doctoral Dissertations and Subsequent Publications that Use NHANES Data. Hairui Yu, S. Perumean-Chaney, K. Kaiser (2023). Journal of Statistics and Data Science Education.