Robust Causal Inference for Large-Scale Healthcare Cost Control and Clinical Efficiency Optimization

DOI: https://doi.org/10.62517/jbdc.202601118

Author(s)

Sining Chai,Tianyu Yang

Affiliation(s)

Northeastern University, Boston, MA, USA

Abstract

To address the core contradiction of "rising costs and insufficient efficiency" in the healthcare system, this study proposes a healthcare system optimization framework integrating robust causal inference and large-scale optimization, targeting issues such as high-dimensional interference, hidden confounding, and lack of model robustness in observational healthcare data. First, a Robust Causal Inference Model (RCI-Model) is constructed, which eliminates hidden confounding through spectral transformation debiasing and identifies core associations via multi-scale causal graph pruning to achieve accurate estimation of clinical causal effects. Furthermore, with causal effects as constraints, an integer programming model incorporating efficacy and resource limitations is established, and cross-institutional resource optimization is completed by combining Lagrangian duality and federated learning. Experiments based on 530,000 hospitalization samples show that the average hospitalization cost is reduced by 18.7%, the bed turnover rate is increased by 23.4%, and the total regional healthcare cost across institutions is reduced by 15.2%. The research confirms that robust causal inference can separate spurious correlations in healthcare data, providing a reliable basis for cost control and efficiency optimization, and technical support for the implementation of hierarchical diagnosis and treatment.

Keywords

Robust Causal Inference; Large-Scale Optimization; Healthcare Cost Control; Clinical Efficiency; Hidden Confounding Elimination

References

[1]Ghosh A, Rothenhäusler D. Assumption-robust Causal Inference[J]. arXiv preprint arXiv:2505.08729, 2025. [2]Cho E, Yang S. Variable selection for doubly robust causal inference[J]. Statistics and its interface, 2024, 18(1): 93. [3]Xiao T, Wang S. Towards unbiased and robust causal ranking for recommender systems[C]//Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining. 2022: 1158-1167. [4]Wang J W, Dai M W, Liang P, et al. Correlation does not equal causation: the imperative of causal inference in machine learning models for immunotherapy[J]. Frontiers in Immunology, 2025, 16: 1630781. [5]Zhang T, Shan H R, Little M A. Causal GraphSAGE: A robust graph method for classification based on causal sampling[J]. Pattern recognition, 2022, 128: 108696. [6]Sanchez P, Voisey J P, Xia T, et al. Causal machine learning for healthcare and precision medicine[J]. Royal Society Open Science, 2022, 9(8): 220638. [7]Zhang Y, Chen S, Liu D. A measurement study of the environmental quality and medical expenditures of elderly individuals: causal inference based on machine learning. Arch Public Health. 2024 Oct 30;82(1):195. [8]Raita Y, Camargo Jr C A, Liang L, et al. Big data, data science, and causal inference: a primer for clinicians[J]. Frontiers in Medicine, 2021, 8: 678047. [9]Hu L, Gu C. Estimation of causal effects of multiple treatments in healthcare database studies with rare outcomes[J]. Health Services and Outcomes Research Methodology, 2021, 21(3): 287-308. [10]Schneeweiss S. Automated data-adaptive analytics for electronic healthcare data to study causal treatment effects[J]. Clinical epidemiology, 2018: 771-788. [11]Irankhah E, Pagare M, Chetla L, et al. Machine learning-enhanced causal inference of surgical decisions and rehabilitation strategies in traumatic brain injury[J]. Frontiers in Neurology, 2025, 16: 1685335.