Tag
This paper presents an AI-augmented decision-support system for ambulance travel time estimation using historical Skåne region data, showing that travel times vary by hour, day, and trip priority while leveraging SHAP and permutation importance for interpretability.
提出一种面向救护车行程时间估计的不确定性感知方法，通过生成上下文自适应的预测区间来支持更透明、更风险敏感的紧急医疗调度决策，并在瑞典Skåne地区两年的真实救护数据上进行了回顾性评估。
This paper investigates systematic pro-AI bias in LLMs, finding that models disproportionately recommend AI-related options in advice-seeking queries, overestimate AI job salaries, and that 'Artificial Intelligence' holds unusually central internal representations regardless of framing, with proprietary models showing stronger bias.
THPL：面向循环水养殖系统（RAS）中虹鳟投喂管理的视觉到语言决策支持框架，通过鱼类轨迹提取行为系数、层级行为编码器生成双证据令牌，并结合LoRA微调与反事实多模态DPO提升LLM的因果推理与决策准确率。
Jev by TypeSafe is a decision component for enterprise AI that provides typed judgments (Choice, Score, Noul) to replace text generation in workflows, enabling better governance and validation.
OpenClaw v2026.9.6 adds support for AI models like Claude Opus 5.5, GPT-6 Sol and Luna, and Grok 4.7, along with features for managed updates, restart recovery, usage reporting, GitHub integration, live meeting notes, and optional decision models.
GLiNER2.5-Decide is a 340M parameter English classification model for operational decisions, supporting multiple label sets in a single forward pass without prompt templates.
This paper investigates offline multimodal large language models as decision support tools for air operations, detailing a modular retrieval-augmented architecture and presenting a pilot study with the Brazilian Air Force that demonstrates reduced cognitive workload and improved efficiency in doctrinal assessment tasks.
This paper proposes a hybrid agentic AI framework for supply chain analytics that uses a coordinator agent and specialized agents to improve decision-making, achieving 90% accuracy and reducing token usage by fourfold.
Phorecaster365 presents a human-supervised reference architecture for pharmaceutical sales forecasting that integrates data ingestion, modeling, and governance with a validation framework, though it does not establish real-world accuracy.
This paper presents the Semantic Signal-Assisted Decision Support (SSADS) framework, which converts return notes into condition factors and signal-quality scores to optimize inspection and recovery allocation in reverse logistics, showing improvements in net recovery value in synthetic benchmark scenarios.
Simile is building a 'What-If Machine' to simulate real-world decisions and understand the impact of interventions, emphasizing causation over correlation.
The paper proposes FLARE, a framework combining fuzzy logic, time-driven activity-based costing, and ROI analysis to assess the economic and operational implications of AI adoption in healthcare under uncertainty.
This paper proposes a Metamorphic Artificial Age Score (AAS) decision-support prototype for flight-log-based drone propeller health monitoring, using a multi-indicator approach to evaluate propeller health and prioritize maintenance.
This paper presents a sociotechnical AI pipeline for ITSM ticket data, combining LLM-based schema normalization and clustering to generate executive-facing decision-support artifacts. Stakeholder evaluation shows strong ratings across interpretability, actionability, trust, and likelihood of use.
This arXiv paper benchmarks eight open-source small language models under different fine-tuning strategies for emergency department decision support, finding that LoRA-tuned SLMs can outperform commercial baselines on triage and referral tasks while remaining locally deployable.
TumorBoard is a multi-agent decision-support system for longitudinal neuro-oncology that uses a shared longitudinal case state and auditable claim-evidence ledger. It outperforms baselines on a 360-case benchmark, with a safety governor reducing harmful recommendations.
This paper presents AWARE-FX, an auditable AI/NLP decision-support system that extracts and scores corporate foreign-exchange hedging disclosures from annual reports, evaluated on 24,909 Hong Kong firm-years with FinBERT, ModernBERT, and Qwen3-8B comparisons.
This paper proposes formulating the bridge between planned tasks and performed actions as a quasi-linear Fisher market, allowing fractional credit assignment. It introduces instruments for conservation and junk filtering, and extends the model with entropy regularization to handle noise, unifying it with optimal transport.
HantaWatch is a federated learning framework for hantavirus genomic surveillance that enables collaborative training of sequence-based models without sharing raw data, integrating k-mer feature extraction and adaptive optimization to support risk screening and expert prioritization.