Date of Award

12-15-2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Science

First Advisor

Daniel Berleant

Abstract

Fraud detection remains a critical challenge across industries such as insurance, healthcare, finance, and government. Global losses from fraud and financial crime are estimated in the trillions annually, including billions in healthcare and insurance fraud alone. While effective for prediction, traditional machine learning methods often lack causal interpretability and struggle to adapt to evolving fraud tactics. This dissertation investigates the application of Double Machine Learning (DML), an emerging causal inference technique, to enhance both the accuracy and interpretability of fraud analytics. The research compares DML against established causal inference approaches, leveraging a meta-learning framework to evaluate model performance on accuracy, computational efficiency, scalability, and interpretability. Specific interpretability metrics (i.e., feature compactness, stability, and entropy) are used to assess the richness of causal insights generated by each method. Results demonstrate that DML provides competitive or superior detection accuracy while offering well-calibrated probability estimates and improved robustness against confounding bias. Comparative analysis highlights DML’s potential not only to identify high-risk fraud cases but also to uncover actionable causal drivers behind fraudulent behaviors. By bridging machine learning and causal inference, this work contributes both a methodological advancement and a practical framework for fraud detection. The findings suggest that DML can significantly improve fraud analytics by enabling scalable, interpretable, and causally grounded detection methods. Implications extend beyond fraud prevention to broader domains where causal understanding and predictive accuracy are equally critical.

Download

Share

COinS