Caviglia Marta, Argyri Katerina D, Athanasiadis Spyridon C, Martignano Maurizio, Gui Daniele, Magalini Sabina, Faccincani Roberto, Cioffi Stefano, Tilsed Jonathan, O’Mara Aoife, Henriksson Linda, Tsekeridou Sofia, Lykokanello Filopoimin, Agarogiannis Evangelos, Forcada Jorge, Rampérez Victor, Antunes Nuno, Rocha da Silva Tiago, Manso Marco, Guerra Barbara, Laist Itamar, Rafalowski Chaim
Institute of Communication and Computer Systems, Zografos, Greece.
Scienze Mediche e Chirurgiche, Università Cattolica del Sacro Cuore, Rome, Italy.
J Med Internet Res. 2025 Apr 10;27:e67318. doi: 10.2196/67318.
In the context of mass casualty incident (MCI) management, artificial intelligence (AI) represents a promising future, offering potential improvements in processes such as triage, decision support, and resource optimization. However, the effectiveness of AI is heavily reliant on the availability of quality data. Currently, MCI data are scarce and difficult to obtain, as critical information regarding patient demographics, vital signs, and treatment responses is often missing or incomplete, particularly in the prehospital setting. Although the NIGHTINGALE (Novel Integrated Toolkit for Enhanced Pre-Hospital Life Support and Triage in Challenging and Large Emergencies) project is actively addressing these challenges by developing a comprehensive toolkit designed to support first responders and enhance data collection during MCIs, significant work remains to ensure the tools are fully operational and can effectively integrate continuous monitoring and data management. To further advance these efforts, we provide a series of recommendation, advocating for increased European Union funding to facilitate the generation of diverse and high-quality datasets essential for training AI models, including the application of transfer learning and the development of tools supporting data collection during MCIs, while fostering continuous collaboration between end users and technical developers. By securing these resources, we can enhance the efficiency and adaptability of AI applications in emergency care, bridging the current data gaps and ultimately improving outcomes during critical situations.
在大规模伤亡事件（MCI）管理的背景下，人工智能（AI）代表着充满希望的未来，有望在诸如分诊、决策支持和资源优化等流程中带来改进。然而，人工智能的有效性在很大程度上依赖于高质量数据的可用性。目前，MCI数据稀缺且难以获取，因为关于患者人口统计学、生命体征和治疗反应的关键信息往往缺失或不完整，尤其是在院前环境中。尽管夜莺（NIGHTINGALE，用于在具有挑战性的大型紧急情况中加强院前生命支持和分诊的新型集成工具包）项目正在通过开发一个全面的工具包来积极应对这些挑战，该工具包旨在支持急救人员并在MCI期间加强数据收集，但仍有大量工作要做，以确保这些工具能够全面运行，并能有效地整合持续监测和数据管理。为了进一步推进这些工作，我们提供了一系列建议，主张增加欧盟资金，以促进生成训练人工智能模型所需的多样且高质量的数据集，包括应用迁移学习以及开发支持在MCI期间进行数据收集的工具，同时促进终端用户与技术开发者之间的持续合作。通过确保这些资源，我们可以提高人工智能在紧急护理中的应用效率和适应性，弥合当前的数据差距，并最终改善危急情况下的结果。
