CFChristopher Fulton

Christopher Fulton

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Christopher Fulton has a total of 6 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    A Fusion Framework for Prospective, Forensic Anomaly Detection in Complex Administrative Panel Data

    by Arvind Sharma, Christopher Fulton, Aleksandar Tomic, Nathaniel Bastian, Lawrence Fulton

    Medicare hospital cost reports constitute one of the largest longitudinal administrative datasets describing the financial and operational behavior of U.S.\ hospitals, yet they remain underutilized for proactive oversight and audit prioritization. This study introduces MOSAIC (Multi-Objective Signal Aggregation for Interpretable Classification), a weakly supervised forensic prioritization framework integrating analytical, behavioral, and model-based anomaly signals through multi-objective optimization. Applied to 83,512 hospital facility-years rom the Medicare Hospital Cost Report Information System (HCRIS) dataset (FY2011--2024), MOSAIC learns signal weights on a pre-pandemic Baseline regime (FY2012--2018) and applies them without modification to all other periods and validation data. MOSAIC achieves a Spearman concordance of $\rho_S = 0.5618$ with a financial distress anchor, compared with a maximum benchmark concordance of $\rho_S = 0.1173$. A signal-exclusion robustness analysis yields $\rho_S = 0.5394$, indicating that performance is not driven by anchor-adjacent variables or any single signal family. External validation using independent datasets of hospital bankruptcies and Medicare enforcement actions confirms that MOSAIC risk stratification anticipates real adverse events under a fully prospective design: risk scores constructed from prior-period ($t-1$) data identify high-risk facilities in subsequent periods ($t$), with approximately 53--54\% of events concentrated in the top two risk quintiles versus a 40\% null expectation. These findings demonstrate that distress-aligned anomaly risk in administrative healthcare data is better captured through domain-constrained multi-signal integration than standalone unsupervised or high-capacity reconstruction models, and that MOSAIC provides a scalable and interpretable framework for data-driven audit prioritization.

  • Open Access

    A Fusion Framework for Prospective, Forensic Anomaly Detection in Complex Administrative Panel Data

    by Arvind Sharma, Christopher Fulton, Aleksandar Tomic, Nathaniel D. Bastian, Lawrence Fulton

    Medicare hospital cost reports constitute one of the largest longitudinal administrative datasets describing the financial and operational behavior of U.S. hospitals, yet they remain underutilized for proactive oversight and audit prioritization. This study introduces MOSAIC (Multi-Objective Signal Aggregation for Interpretable Classification), a weakly supervised forensic prioritization framework integrating analytical, behavioral, and model-based anomaly signals through multi-objective optimization. Applied to 83,512 hospital facility-years rom the Medicare Hospital Cost Report Information System (HCRIS) dataset (FY2011--2024), MOSAIC learns signal weights on a pre-pandemic Baseline regime (FY2012--2018) and applies them without modification to all other periods and validation data. MOSAIC achieves a Spearman concordance of rho = 0.5618 with a financial distress anchor, compared with a maximum benchmark concordance of rho = 0.1173$. A signal-exclusion robustness analysis yields rho = 0.5394$, indicating that performance is not driven by anchor-adjacent variables or any single signal family. External validation using independent datasets of hospital bankruptcies and Medicare enforcement actions confirms that MOSAIC risk stratification anticipates real adverse events under a fully prospective design: risk scores constructed from prior-period (t-1) data identify high-risk facilities in subsequent periods ($t$), with approximately 53--54\% of events concentrated in the top two risk quintiles versus a 40\% null expectation. These findings demonstrate that distress-aligned anomaly risk in administrative healthcare data is better captured through domain-constrained multi-signal integration than standalone unsupervised or high-capacity reconstruction models, and that MOSAIC provides a scalable and interpretable framework for data-driven audit prioritization.