Multi-criteria models in healthcare operations management require weights that reflect how stakeholders trade off patient waiting time, provider workload, and operating cost. According to a new arXiv preprint, these weights are difficult to quantify, which makes it hard to apply such models in practice.
The paper investigates whether teams of large language models—multi-agent systems—can help address this challenge. The abstract frames the work as an exploration of both possibilities and limitations, suggesting that while LLM agents may offer a novel way to elicit or approximate stakeholder preferences, their use is not without constraints.
Because the abstract is brief and does not include results, the article does not report specific findings. The contribution at this stage is the framing of the problem and the proposed multi-agent approach for a domain where quantitative trade-offs are inherently subjective.