Causal reinforcement learning to optimise hospital bed allocation under surge conditions
NexusTech · 09.04.2026 · 1 (2) · 2026006
DOI: 10.31893/tech.2026006 ↗Kurz gefasst
Fachartikel in der Zeitschrift NexusTech, online erschienen am 09.04.2026. Verfasst von Akhilesh Kumar. Die Arbeit nennt 15 Literaturangaben.
Abstract (Originalsprache)
Hospital systems globally face extreme operational pressure during surge conditions where demand exceeds capacity. This study proposes a Causal Reinforcement Learning (CRL) framework integrating Structural Causal Models with Deep Q-Networks to optimize hospital bed allocation. Counterfactual reasoning improves decision-making under uncertainty. A discrete-event simulation of a 500-bed hospital was conducted over 10,000 episodes under varying surge intensities. Results demonstrate a 14.5% reduction in mortality, a 28.4% reduction in boarding time, and improved resource utilization. Statistical validation using Wilcoxon tests confirms significance (p < 0.01). The framework also enhances interpretability via SHAP-based causal attribution. The proposed CRL approach provides a scalable, interpretable, and robust solution for surge-aware healthcare resource optimization, particularly in resource-constrained environments.
Zitieren
APA
Kumar, A. (2026). Causal reinforcement learning to optimise hospital bed allocation under surge conditions. NexusTech, 1(2), 2026006. https://doi.org/10.31893/tech.2026006
Vancouver
Kumar A. Causal reinforcement learning to optimise hospital bed allocation under surge conditions. NexusTech. 2026;1(2):2026006. doi:10.31893/tech.2026006
BibTeX
@article{Kumar2026,
author = {Kumar, Akhilesh},
title = {Causal reinforcement learning to optimise hospital bed allocation under surge conditions},
journal = {NexusTech},
year = {2026},
volume = {1},
number = {2},
pages = {2026006},
doi = {10.31893/tech.2026006}
}
EUVIMED