Three New arXiv Papers Explore LLM Enhancements for Forecasting, Decision-Making, and Medical Applications

Three New arXiv Papers Explore LLM Enhancements for Forecasting, Decision-Making, and Medical Applications

Three recent papers on arXiv address different challenges in applying Large Language Models to specialized domains.

STELLA for Time Series Forecasting

According to arXiv paper cs.AI/2512.04871v1, current adaptations of LLMs for time series forecasting “often fail to effectively enhance information for raw series, leaving LLM reasoning capabilities underutilized.” The paper notes that existing prompting strategies rely on static correlations, though the abstract provided is incomplete.

What-If Analysis for Game Decision-Making

A paper (arXiv:2509.04791v2) examines LLM limitations in dynamic environments like MOBA games. According to the abstract, “Large Language Models (LLMs) are effective at reasoning and information retrieval, but remain unreliable for decision-making in dynamic, partially observable, high-stakes environments.” The research identifies “weak counterfactual reasoning” as a key limitation.

Medical Practice Alignment

The third paper (arXiv:2511.16139v3) addresses challenges in integrating LLMs into medical practice. According to the abstract, the research identifies “a misalignment between static evaluation benchmarks” and real-world clinical needs as constraining factors, though the abstract excerpt is incomplete.

All three papers represent ongoing research efforts to address specific limitations in LLM applications across different domains.